PDF Communications in Interference Limited Networks 1St Edition Wolfgang Utschick Eds Ebook Full Chapter

Communications in Interference
Limited Networks 1st Edition Wolfgang

Utschick (Eds.)
Visit to download the full and correct content document:
https://textbookfull.com/product/communications-in-interference-limited-networks-1st-
edition-wolfgang-utschick-eds/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Networks and Communications NetCom2013 Proceedings of

the Fifth International Conference on Networks
Communications 1st Edition Natarajan Meghanathan
https://textbookfull.com/product/networks-and-communications-
netcom2013-proceedings-of-the-fifth-international-conference-on-
networks-communications-1st-edition-natarajan-meghanathan/
Full Duplex Communications and Networks 1st Edition

Lingyang Song
https://textbookfull.com/product/full-duplex-communications-and-
networks-1st-edition-lingyang-song/
Security Aware Device to Device Communications

Underlaying Cellular Networks 1st Edition Aiqing Zhang
https://textbookfull.com/product/security-aware-device-to-device-
communications-underlaying-cellular-networks-1st-edition-aiqing-
zhang/
Intelligent Vehicular Networks and Communications.

Fundamentals, Architectures and Solutions 1st Edition
Anand Paul
https://textbookfull.com/product/intelligent-vehicular-networks-
and-communications-fundamentals-architectures-and-solutions-1st-
edition-anand-paul/
LTE communications and networks : femtocells and
antenna design challenges First Edition Safdar
https://textbookfull.com/product/lte-communications-and-networks-
femtocells-and-antenna-design-challenges-first-edition-safdar/
Machine Learning and Cognitive Computing for Mobile

Communications and Wireless Networks 1st Edition
Krishna Kant Singh
https://textbookfull.com/product/machine-learning-and-cognitive-
computing-for-mobile-communications-and-wireless-networks-1st-
edition-krishna-kant-singh/
Interference Analysis: Modelling Radio Systems for

Spectrum Management 1st Edition John Pahl
https://textbookfull.com/product/interference-analysis-modelling-
radio-systems-for-spectrum-management-1st-edition-john-pahl/
Big Data Computing and Communications Second

International Conference BigCom 2016 Shenyang China
July 29 31 2016 Proceedings 1st Edition Yu Wang
https://textbookfull.com/product/big-data-computing-and-
communications-second-international-conference-
bigcom-2016-shenyang-china-july-29-31-2016-proceedings-1st-
edition-yu-wang/
Information and Communications Security 18th

International Conference ICICS 2016 Singapore Singapore
November 29 December 2 2016 Proceedings 1st Edition
Kwok-Yan Lam
https://textbookfull.com/product/information-and-communications-
security-18th-international-conference-icics-2016-singapore-
singapore-november-29-december-2-2016-proceedings-1st-edition-
Signals and Communication Technology
Wolfgang Utschick Editor
Communications
in Interference
Limited
Networks
More information about this series at http://www.springer.com/series/4748
Wolfgang Utschick
Editor
Communications
in Interference Limited
Networks
123
Editor
Wolfgang Utschick
Technische Universität München
Munich
Germany
ISSN 1860-4862 ISSN 1860-4870 (electronic)

ISBN 978-3-319-22439-8 ISBN 978-3-319-22440-4 (eBook)
DOI 10.1007/978-3-319-22440-4
Library of Congress Control Number: 2015950002
Springer Cham Heidelberg New York Dordrecht London

© Springer International Publishing Switzerland 2016
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein or
for any errors or omissions that may have been made.
Printed on acid-free paper
Springer International Publishing AG Switzerland is part of Springer Science+Business Media

(www.springer.com)
To Ralf Kötter
Preface
The classical approaches to cope with interference in wireless communication

systems are based on the principle of interference avoidance and minimization. This
is classically achieved through proper network planning as well as various
orthogonal medium access techniques. While network planning is largely con-
cerned with the geographic situation of a given network, relying on interference
control through attenuation of wireless signals, medium access techniques exploit
orthogonality in the signals themselves to keep different users separate. However, as
the demands on modern wireless communication systems increase dramatically,
interference still stands to become the limiting performance factor. Orthogonality,
either induced by time, frequency, or space, cannot be achieved perfectly but can, at
best, offer to mitigate interference. In addition to this imperfect interference control,
it is known that orthogonal access techniques are suboptimal with respect to the
fundamentally achievable rates in a network. In contrast to and beyond orthogonal
access techniques, centralized and decentralized methods have been proposed to
avoid and handle interference as well as approaches that resolve interference
constructively. In fact, an exiting insight into the operation of networks is that it
may be beneficial, in terms of an overall throughput, to actively create and manage
interference. Thus, when handled properly, “mixing” of data in networks becomes a
useful tool for operation rather than a nuisance, as it has been treated traditionally.
The development of mobile, robust, ubiquitous, reliable, and instantaneous com-
munication being a driving and enabling factor of an information centric economy,
the understanding, mitigation, and exploitation of interference in networks must be
seen as a centrally important task.
This book exemplarily introduces the fundamentals of various approaches to aim
at these objectives. It summarizes and extends the collective scientific work that has
been conducted under the umbrella of a priority program by the same name that has
been supported by the Deutsche Forschungsgesellschaft (DFG) from 2009 to 2015.
The chapters in this book are related to a wide range of challenges and topics, among
which are ad hoc networks, analog computation in networks, cellular networks,
cooperative communications, device-to-device communications, evidence-based
vii
viii Preface
computation models, interference alignment, interference limited networks, inter-

ference management, mesh networks, multicasting in networks, multiple-relay
networks, network coding, physical layer security, robust communications, satellite
networks, and many more. With this book the authors provide an extensive and
substantial overview of current research in a buoyant technological area for future
wireless communication systems.
The idea of this book was born during a series of meetings over the span of the
DFG priority program Communications in Interference Limited Networks that was
started with a joint initiative of a group of researchers in 2009. My special thanks go to
Holger Boche (Technische Universität München), Robert Fischer (Universität Ulm),
Peter Höher (Christian-Albrechts-Universität zu Kiel), Anja Klein (Technische
Universität Darmstadt), Volker Kühn (Universität Rostock), Berthold Lankl
(Universität der Bundeswehr München), and Erich Lutz (Deutsches Zentrum für
Luft- und Raumfahrt). My special thanks also to my colleagues in the coordination
group, Johannes Huber (Friedrich-Alexander-Universität Erlangen-Nürnberg), and
Ralf Kötter (Technische Universität München). In particular, I want to thank Ralf
Kötter (2009y) for all the fun during preparing the proposal of the program and the
unforgettable time with him at the Technische Universität München. It was a great
tragedy that Ralf could not participate in the program any more.
Finally, I want to thank all authors and co-authors who contributed to this book,
and Christoph Baumann, Gayathri Umashankar, and Mathini Manoj from Springer
International Publishing AG, and last but not least, Thomas Wiese, Technische
Universität München, who did an incredible job of supporting the collection and
harmonization of all material.
Munich, Germany Wolfgang Utschick

July 2015
Contents
1 Coding Techniques for Transmitting Packets Through Complex

Communication Networks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Martin Bossert, Vladimir Sidorenko and Antonia Wachter-Zeh
2 Modulo-Type Precoding for Networks. . . . . . . . . . . . . . . . . . . . . 31
Robert F.H. Fischer, Michael Cyran, Sebastian Stern
and Johannes B. Huber
3 Enabling the Multi-User Generalized Degrees of Freedom in
Cellular Interference Networks with Multi-User Coding. . . . . . . . 53
Rick Fritschek and Gerhard Wunder
4 The Information-Theoretic Constant-Gap Optimality of Treating
Interference as Noise in Interference Networks . . . . . . . . . . . . . . 75
Soheil Gherekhloo, Anas Chaaban and Aydin Sezgin
5 Interference-Aware Analog Computation over the Wireless
Channel: Fundamentals and Strategies . . . . . . . . . . . . . . . . . . . . 97
Mario Goldenbaum, Sławomir Stańczak and Holger Boche
6 Bounds on the Outage Constrained Capacity of the Gaussian
Relay Channel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
Andreas Gründinger, Lennart Gerdes, Michael Joham
and Wolfgang Utschick
7 Balancing for Interference-Limited Multi-User Satellite
Communications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
Andreas Gründinger, Michael Joham and Wolfgang Utschick
ix
x Contents
8 Distributed Channel Selection for Underlay Device-to-Device

Communications: A Game-Theoretical Learning Framework . . . . 173
Setareh Maghsudi and Sławomir Stańczak
9 Cyclic Interference Alignment Via Polynomial Rings for
Multi-user Communication Channels . . . . . . . . . . . . . . . . . . . . . 191
Henning Maier, Johannes Schmitz and Rudolf Mathar
10 Multicast in Networks of Broadcast Channels—Part I:
Submodular Models and Optimization . . . . . . . . . . . . . . . . . . . . 215
Maximilian Riemensberger and Wolfgang Utschick
11 Multicast in Networks of Broadcast Channels—Part II:
Representation of Bounds on the Multicast Capacity Region . . . . 237
Maximilian Riemensberger and Wolfgang Utschick
12 Arbitrarily Varying Channels—A Model for Robust
Communication in the Presence of Unknown Interference . . . . . . 259
Rafael F. Schaefer, Holger Boche and H. Vincent Poor
13 Resource Allocation and Pricing in Non-cooperative Interference
Networks with Malicious Users . . . . . . . . . . . . . . . . . . . . . . . . . . 285
Fei Shen, Anil Kumar Chorppath, Eduard Jorswieck
and Holger Boche
14 Interference Alignment Aided by Non-Regenerative Relays . . . . . 327
Rakash SivaSiva Ganesan, Hussein Al-Shatri, Xiang Li, Anja Klein
and Tobias Weber
15 Stochastic Geometry for Analysis of Coordination and
Cooperation in Interference–Limited Ad Hoc Networks . . . . . . . . 347
Andrey Skrebtsov, Guido H. Bruck and Peter Jung
16 Cooperative Diversity Under Spatial Interference Correlation
in Wireless Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
Ralph Tanbourgi, Holger Jäkel and Friedrich K. Jondral
17 Physical-Layer Key Generation and Reconciliation . . . . . . . . . . . 393
Jon Wallace, Rashid Mehmood, Rajesh Sharma,
Werner Henkel, Oana Graur, Nazia Islam and Alexandra Filip
18 Physical Layer Cooperation in One-Way Relaying Systems . . . . . 431
Meng Wu, Frank Ludwig, Dirk Wübben, Armin Dekorsy,
Karl-Dirk Kammeyer and Steffen Paul
Contents xi
19 Physical Layer Cooperation in Two-Way Relaying Systems . . . . . 451

Meng Wu, Frank Ludwig, Dirk Wübben, Armin Dekorsy
and Steffen Paul
20 Generalized Multi-Carrier Waveforms in Two-Way Relay
Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 475
Matthias Woltering, Stephan Schedler, Dirk Wübben, Armin Dekorsy
and Volker Kühn
21 Linear Multi-Cell Precoding for Throughput Optimization
Considering Outage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501
Xin Zhang, Richard Fritzsche, Andreas Festag and Gerhard Fettweis
Chapter 1
Coding Techniques for Transmitting
Packets Through Complex
Communication Networks
Martin Bossert, Vladimir Sidorenko and Antonia Wachter-Zeh
Abstract Random Linear Network Coding (RLNC) is a technique to disseminate

information in a network. Various error scenarios require algebraic code construc-
tions with high error-correcting capability in order to transmit packets reliably
through such a network. It was shown that subspace codes, in particular lifted rank-
metric codes, are suitable for this purpose, in contrast to Hamming metric in the
case of a classical transmission. The mainly used codes are Gabidulin codes. In this
contribution, we will introduce Gabidulin codes and describe several error-erasure
decoding algorithms. Further, an extension of Gabidulin codes is introduced which
allows to decode beyond half the minimum rank distance, the interleaved Gabidulin
codes. Further, we will introduce (partial) unit memory codes based on Gabidulin
codes. Such convolutional codes are of particular interest in so-called multi-shot
transmissions since memory between different transmission is introduced. Finally,
we will show a significant difference of Gabidulin and Reed-Solomon codes in case
of list decoding. Namely, that the list size can grow exponentially for a decoding
radius below the Johnson bound for rank metric codes.
1.1 Introduction
Random linear network coding (RLNC) is a technique for transmitting packets

through complex communication networks. In Fig. 1.1, such a network is depicted.
M. Bossert (B) · A. Wachter-Zeh

Ulm University, Albert-Einstein-Allee 43, 89081 Ulm, Germany
e-mail: martin.bossert@uni-ulm.de
V. Sidorenko
Institute for Information Transmission Problems,
Russian Academy of Sciences, Moscow, Russia
e-mail: vladimir.sidorenko@uni-ulm.de
A. Wachter-Zeh
Technion—Israel Institute of Technology, Haifa, Israel
e-mail: antonia@cs.technion.ac.il
© Springer International Publishing Switzerland 2016 1

W. Utschick (ed.), Communications in Interference Limited Networks,
Signals and Communication Technology, DOI 10.1007/978-3-319-22440-4_1
2 M. Bossert et al.
Fig. 1.1 Random linear network [4]
We want to transmit information through a network from a source to a destination.

In RLNC, the sent packets are (n + m)-dimensional vectors over a finite field1 Fq
and each node calculates its outgoing packets as random linear Fq -combinations
of its incoming packets. The transmission takes place in shots, where in each shot
the source inserts n packets x0 , . . . , xn−1 (rows of a matrix X ∈ Fqn×(n+m) ) into
the network and the destination collects ñ ≥ n packets y0 , . . . , yñ−1 (rows of a
matrix Y ∈ Fqñ×(n+m) ). If the transmission is error-free, the collected packets at the
destination are random linear combinations of the transmitted vectors, which implies
that with high probability the row spaces of X and Y are the same. In [20], the row
space of X is seen as the message. Due to the linear combinations of the network,
we get:
Y = AX,
where the matrix A ∈ Fqñ×n describes the transformation of the network. If nodes
perform a random linear combination of incoming packets during one shot, random
additive errors within the network can be described as
Y = AX + BZ,
where B ∈ Fqñ× represents the linear transformation of the network and the rows
of Z ∈ Fq×(n+m) are the error vectors z0 , . . . , zñ−1 of the corresponding edge of the
1F denotes a finite field with q elements, where q is a prime power.

q
1 Coding Techniques for Transmitting Packets … 3
network with elements from Fq . E.g, zi = 0 means that on the edge corresponding
to the ith row of B there did not occur any error.
Note that the only property which is not changed by a transmission over an
error-free random linear network coding channel is the row space of the transmit-
ted matrix X. Therefore, information is encoded into subspaces instead of specific
matrices.
In [36], it is shown that subspace codes and rank-metric codes are closely related.
In [36], a construction is given where Gabidulin codes are used for error control with
respect to errors and erasures. We transmit a codeword

X = In M ,
where M is a (n × m) code matrix of a matrix code C M and In is the n × n identity

matrix. In [36], the use of Gabidulin codes in rank-metric is proposed. It follows that
Y = AX + Eout ,
is received, where Eout is the error matrix. Now, let S be a matrix which brings Y to
echelon form. Then, we have

GR
SY = .
0 C
The matrix R has r rows such that C consists of ñ − r rows. The paper [36] extended
R to n rows by inserting specific zero rows to obtain R̂ and showed that R̂ can be
decomposed into
R̂ = M + LM + DC + Er est
= M + Er ow + Ecol + Er est ,
where M is the transmitted codeword. The matrix L is a known (n × n) matrix such

that LM is a generalized rank row erasure. In [36], this is called an erasure. The
(ñ − r × m) matrix C is known as well and DC is a generalized rank column erasure
which is called deviation in [36]. The remaining Er est is a random rank error. This
decomposition will not be taken into consideration since we do not regard erasures.
We will only discuss Y = AX + Eout .
Clearly, the matrix A can be used to check whether with row operations the identity
matrix can be recovered. For simplicity, the case where A does not have full rank
will not be considered in the following. Then, the Gabidulin code must be decoded
in order to correct errors. Therefore, in the next section we describe Gabidulin codes
and their decoding.
4 M. Bossert et al.
1.2 Gabidulin Codes
While in [5] network coding with error correction is studied for the operator channel,
in [16] the benefits of network coding over routing in random networks are discussed.
In [35], codes for network coding have been introduced, which are based on rank-
metric codes, in particular Gabidulin codes. Since these codes are not commonly
known, we will briefly describe them in the following.
The notion rank norm outside of coding theory is known (implicitly) in linear
algebra by the well known inequality
rk(A + B) ≤ rk(A) + rk(B)
Explicitly, the rank norm was introduced by Hua [17]. Later, it was introduced to
coding theory in [6, 11, 29].
For a given basis of Fq m over Fq , there exists a vector space isomorphism which
maps each vector x ∈ Fqn m to a matrix X ∈ Fqm×n . Let rk(x) denote the rank of X over
Fq and let Rq (X), Cq (X) denote the row and column space of X in Fqn . We use the
notation as vector (e.g. from Fqn m ) or matrix (e.g. from Fqm×n ) equivalently, whatever
is more convenient. The rank distance of two vectors x, y ∈ Fq m is defined by
dR (x, y) = rk(x − y)
and we define the minimum rank distance d of a code C as

d = min dR (c1 , c2 ) : c1 , c2 ∈ C, c1 = c2 .
An Cn,M,d code over Fq m denotes a code (not necessarily linear) of cardinality M

and minimum rank distance d. Its codewords are in Fqn m or equivalently represented
as matrices in Fqm×n . W.l.o.g. we assume in this paper that n ≤ m. If this is not the
case, we consider the transpose of all matrices such that n ≤ m holds. We call n the
length of such a block code in rank metric over Fq m .
The cardinality M of an Cn,M,d code with n ≤ m is limited by a Singleton-like
upper bound (see [11, 36]):
M ≤ q min{n(m−d+1),m(n−d+1)} = q m(n−d+1) . (1.1)
For linear codes of length n ≤ mGabidulin and dimension k, this implies d ≤

n − k + 1. If the cardinality of a code fulfills (1.1) with equality, the code is called
a Maximum Rank Distance (MRD) code. A linear MRD code over Fq m of length
n ≤ m, dimension k and minimum rank distance d = n − k + 1 is denoted by
MRD[n, k] and has cardinality M = q mk . In the following we introduce the class of
Gabidulin codes which are are MRD codes.
The codes introduced in [11] have length n, dimension k, minimum rank distance
d = n − k + 1, and are nowadays called Gabidulin codes. Gabidulin codes are
constructed over an extension field Fq m of Fq . A generator matrix is given by:
⎡ ⎤
g1 g2 · · · gn
⎢ g [1] g2[1] · · · gn[1] ⎥
⎢ 1 ⎥
⎢ [2] ⎥
G=⎢
⎢ g1 g2[2] · · · gn[2] ⎥,
⎥ (1.2)
⎢ .. .. . . .. ⎥
⎣. . .. ⎦
g1[k−1] g2[k−1]
· · · gn[k−1]
where g j ∈ Fq m , j = 1, . . . , n, are linearly independent over Fq . The notation

g [i] = g q denotes the i-th Frobenius power of g.
i
A parity-check matrix has the form

⎡ ⎤
h1 h2 . . . hn
⎢ h [1] h [1] . . . h [1] ⎥
⎢ 1 2 n ⎥
H=⎢ .. .. .. ⎥ , (1.3)
⎣ . . ... . ⎦
h 1[d−2] [d−2]
h2 . . . h n[d−2]
where h j ∈ Fq m , j = 1, . . . , n, are linearly independent over Fq , and
GH = 0.
Gabidulin codes are the rank-metric analogs of Reed-Solomon codes and found
many applications including network coding. Interleaving or the direct sum of
Gabidulin codes allows both decreasing the redundancy and increasing the error cor-
recting capability for network coding. For Gabidulin codes, we proposed a transform-
domain algorithm correcting both errors and erasures [32]. We showed how to gener-
alize this algorithm for interleaved Gabidulin codes. The transform-domain approach
allows to simplify derivations and proofs and also simplifies finding the error vector
after solving the key equation. Details on these results can be found in [33].
The classical definition of Gabidulin codes requires the concept of linearized
polynomials. A linearized polynomial (or q-polynomial) over Fq m is a polynomial
of the form
t
f (x) = f i x [i]
i=0
where f i ∈ Fq m . The maximum i for which f i = 0 is called the q-degree of f (x)

and is denoted by degq f (x). The addition of linearized polynomials f (x) and g(x)
is defined componentwise. Multiplication uses the symbolic product: f (x) ◦ g(x) =
f (g(x)). The result is a linearized polynomial of q-degree degq f (x) + degq g(x).
6 M. Bossert et al.
The symbolic product is distributive and associative, but noncommutative. The set
of linearized polynomials over Fq m with the operation of polynomial addition and
symbolic multiplication forms a noncommutative ring.
1.2.1 Decoding Using the Linearized Extended Euclidean

Algorithm (LEEA)
There exist several efficient algorithms for decoding Gabidulin codes. A syndrome-
based bounded minimum distance (BMD) decoding algorithm based on solving
a key equation was introduced by Gabidulin [11]. The first step of the decoding
process is to compute the syndrome. This can be done by multiplying the received
word r with the transposed of the parity check matrix H. We use the syndrome
values s = (s0 , . . . , sn−k+1 ) := r · HT to calculate the syndrome polynomial s(x) =
n−k−1
i=0 si x i . The key equation which has to be solved is given by
Ω(x) ≡ Λ(s(x)) = Λ(x) ◦ s(x) mod x [n−k] , deg Ω(x) < deg Λ(x).
This key equation can be solved for the error span polynomial Λ(x). The direct way
to accomplish this task is to solve a linear system of equations based on the key
equation:

i
[ j]

t
[ j]
Ωi = Λ j si− j = Λ j si− j = 0, ∀i ∈ [t, n − k − 1 − 1].
j=0 j=0
This is equivalent to the following homogeneous linear system of equations:

⎛ ⎞ ⎛ ⎞ ⎛ ⎞
Ωt st[0] [1]
st−1 . . . s0[t] Λ0
⎜ Ωt+1⎟ ⎜ [0]
st+1 st[1] . . . s1[t] ⎟ ⎜Λ1 ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎜ ..⎟=⎜ .. .. .. .. ⎟ · ⎜ .. ⎟ = 0. (1.4)
⎝ .⎠ ⎝ . . . . ⎠ ⎝ . ⎠
Ωn−k−1 [0] [1] [t] Λt
sn−k−1 sn−k−2 . . . sn−k−1−t
If we use Gaussian elimination, the complexity of this step is bounded by O(n 3 )

operations in Fq m . However, it is possible to exploit the structure of the syndrome
matrix in order to solve the key equation with complexity O(n 2 ) in Fq m by applying
the Linearized Extended Euclidean Algorithm (LEEA) which is given in Algorithm 1.
The algorithm uses a subroutine RightDiv(a(x); b(x)) which finds two polynomials
q(x) and r (x) such that
a(x) = q(x) ◦ b(x) + r (x),
and degq r (x) < degq b(x). The existence of such an algorithm is ensured by the
fact that the ring of linearized polynomials with addition + and multiplication ◦ is a
Euclidean domain.
Algorithm 1:
rout (x); u out (x); vout (x) ← RightLEEA a(x); b(x); dstop
Input: a(x); b(x) ∈ Lq m [x] with degq a(x) ≥ degq b(x);
stopping degree dstop
Initialize: i ← 1,
r (−1) (x) ← a(x), r (0) (x) ← b(x),
u (−1) (x) ← 0, u (0) (x) ← x [0] ,
v (−1) (x) ← x [0] , v (0) (x) ← 0
1 while degq r (i−1) (x) ≥ dstop do

2 q (i) (x); r (i) (x) ← RightDiv r (i−1) (x); r (i−2) (x) u (i) (x) ← u (i−2) (x) − q (i) (u (i−1) (x))
v (i) (x) ← v (i−2) (x) − q (i) (v (i−1) (x)) i ← i + 1
Output: rout (x) ← r (i−1) (x); u out (x) ← u (i−1) (x); vout (x) ← v (i−1) (x)
Therefore, we use the syndrome polynomial s(x) and x [n−k] as input to the LEEA
which in turn gives us Λ(x). Next, we need to find a basis of the root space of
Λ(x). This is relatively easy due to the structure of linearized polynomials. This
basis can be used for determining the error. Therefore it is necessary to solve a
system of equations. The overall complexity is in the order O(n 2 ) over Fq m . We
can further reduce the complexity by applying a Gao-like decoding approach [47].
This algorithm uses an alternative transformed key equation which can be solved by
applying the LEEA and outputs directly the linearized evaluation polynomial of the
estimated codeword.
1.2.2 Linearized Shift-Register Synthesis
Another algorithm that can be applied for solving the key equation efficiently is a gen-
eralization of the Berlekamp-Massey algorithm [31]. A problem which is equivalent
to the problem of solving the key equation is the problem of linearized shift-register
synthesis: Let s = s1 , s2 , . . . , s N be a sequence over a field Fq m . Find the smallest
integer l ≥ 0 for which there is a vector σ = (σ1 , . . . , σl ) over Fq m such that

l
qi
sn = − σi sn−i f or n = l + 1, . . . , N
i=1
8 M. Bossert et al.
and find a suitable vector σ which is called the connection vector. Given a connection
vector σ = (σ1 , . . . , σl ) over Fq m , the corresponding connection q-polynomial is
defined by
l
σ (x) = σi x [i] , (1.5)
i=0
Δ
where σ0 = 1 and degq σ (x) ≤ l. A linearized-feedback shift-register is com-
pletely determined by length l and connection vector σ , hence it can be denoted by
(l, σ ). Given a sequence s and a shift-register (l, σ ), let the discrepancy dn (l, σ ) of
a sequence element sn be defined by
⎧
⎨0 for n = 1, . . . , l,
dn (l, σ ) = l
[i] (1.6)
⎩ σi sn−i for n = l + 1, . . . , N .
i=0
A shift-register (l, σ ) generates the element sn , if dn (l, σ ) = 0. The shift-register

generates a sequence if it generates every element of the sequence.
Algorithm 2:
l; σ (x); σ (x); n ← Richter–Plass(s)
Input: s = s1 , . . . , s N
Initialize: l ← 0, σ (x) ← x
σ (x) ← x, n ← 0, d ← 1
3 for each n from 1 to N do
[ j]
4 d ← lj=0 σ j sn− j if d = 0 then
5 if 2l ≥ n then
[n−n ]
6 σ (x) ← σ (x) − d dx ⊗ σ (x)
7 else
x [n−n ]
8 σ̃ (x) ← σ (x) σ (x) ← σ (x) − d d ⊗ σ (x) l ← n − l σ (x) ← σ̃ (x),
n ← n, d ← d
Output: l, σ (x), and σ (x), n
The problem of linearized shift-register synthesis can be tackled by solving a linear

system of equations. Applying straightforward approaches like Gaussian elimination
have a computational complexity of order O(N 3 ). Solving the problem with Algo-
rithm 2 reduces the complexity to order O(N 2 ). The algorithm was independently
introduced by Paramonov and Tretjakov [26] and by Richter and Plass [28].
Algorithm 2 is an iterative approach. First, the algorithm initializes the setting
of the shift-register with (l, σ ) = (0, x). The for-loop sequentially processes the
elements of s. Within the loop, the discrepancy dn (l, σ ) is computed using the cur-
rent shift-register (l, σ ). The discrepancy is zero if and only if (l, σ ) is able to generate
the element sn . In this case there is no need to change the shift-register. Hence, the
algorithm directly continues with the processing of the next element. If the discrep-
ancy is not equal to zero, the shift-register (l, σ ) is modified in order to obtain the
shortest shift-register (l,ˆ σ̂ ) which is able to generate sn and all previous elements.
The algorithm outputs both, length l and a connection polynomial σ .
The time complexity of this algorithm may be estimated by O(l N ) ≤ O(N 2 )
operations in Fq m since the algorithm has to process N elements of the sequence s
and the processing of each element requires at most O(l) ≤ O(N ) operations in Fq m .
In addition, the uniqueness of the solution can be checked. For decoding Gabidulin
codes up to half the code distance, a decoding failure can be declared when the
solution is not unique. Furthermore, it is possible to calculate all solutions which is
interesting for decoding Gabidulin codes beyond half the code distance.
1.2.3 Remarks
Fast decoding up to half the minimum rank distance of Gabidulin codes by applying a
Gao-like algorithm is explained in [47]. Sidorenko et al. [31] introduces the general-
ization of the Berlekamp-Massey algorithm and proves its correctness. Furthermore
it is explained how to check uniqueness of the solution and how to find all possible
solutions.
1.3 Interleaved Gabidulin Codes
Interleaving or the direct sum of Gabidulin codes with rank distance d can be
applied for random network coding. For codes over Fq m , a fast decoding algorithm

exists which corrects errors of rank up to +1 (d − 1) with high probability. Thus,
with interleaving the error correcting radius can be enlarged. The algorithm is based
on fast skew-feedback shift-register synthesis and for fixed it has time complexity
O(m 3 log m) operations in the base field Fq .
In network coding, rows of a matrix C ∈ Fqm×n (with a prefix of length m) are
sent via the network as packets. In many practical cases the packet length is much
more than the number m of packets in the message, so we would like to have n m
in this case. In order to increase the parameter n by a factor of ∈ N, the direct sum
of codewords of the code C can be used. Given a matrix code C ⊆ Fqm×n and a
number , the -interleaved code IC ⊆ Fqm×n consists of all m × n matrices C over
Fq having the following block form

C = C(1) C(2) . . . C() , C() ∈ C, (1.7)
10 M. Bossert et al.
where l = 1, . . . , . The rate and the distance of the interleaved code are the same
as for the original code C.
We assume m = n for the interleaved Gabidulin code. Note, that Loidreau and
Overbeck in [24, 25] used interleaving of transposed Gabidulin codes.
We describe a decoder for any -interleaved Gabidulin code IC ⊆ Fqm×n having
distance d. This decoder guarantees to correct all error words of rank less than d/2
and it corrects with high probability all error words of rank t if

t≤ (d − 1).
+1
The probability Pf (t) of a decoding failure can be upper bounded by Pf (t) < 4/q m
for the cases that words of fixed rank are equiprobable. This probability is very small
in practice and coincides with the bound from [24, 25] and Pf (t) practically vanishes

if t < +1 (d − 1).
1.3.1 Decoding Interleaved Gabidulin Codes
Consider interleaving of identical Gabidulin codes G(q m ; n, k) and index runs

from 1 to , i.e., = 1, . . . , .
Definition 1.3.1 [32] The -interleaved Gabidulin code is

(1)
IG(q m , ; n, k) = c . . . c() : c(l) ∈ G(q m ; n, k) . (1.8)

Here c = c(1) . . . c() ∈ Fqnm denotes the concatenation of vectors c() . The

codeword c ∈ IG(q m , ; n, k) is transmitted and a vector r = r(1) . . . r() ∈ Fqnm

is received. The error on the channel is e = e(1) . . . e() ∈ Fqnm , where r = c + e.
Let rk(e) = t then we can represent the error vector as
e = aB, where a ∈ Fqt m , B ∈ Fqt×n ,
and rk(a) = rk(B) = t. We can write B as

B = B(1) , . . . , B() , B(l) ∈ Fqt×n ,
where rk(B(l) ) ≤ t. Then we have
e(l) = aB(l) (1.9)

and the syndrome of the l-th code is
s(l) = aB(l) HT . (1.10)

[i]
We use the notation x (l) = x (l)[i] . let
⎛ ⎞
f 1(l) . . . f 1(l)[d−2]
⎜ f 2(l) . . . f 2(l)[d−2] ⎟
⎜ ⎟
F(l) = B(l) HT = ⎜ .. .. .. ⎟, (1.11)
⎝ . . . ⎠
f t(l) . . . f t(l)[d−2]
where

n
f i(l) = Bi,(l)j h j , i = 1, . . . , t. (1.12)
j=1
Thus we have the following system of equations with unknowns ai , f i(l) , i =

1, . . . , t,
t
(l)[ j−1]
ai f i = s (l)
j , j = 1, . . . , d − 1. (1.13)
i=1
We introduce again a q-linearized polynomial

t
σ (x) = σi x [i] (1.14)
i=0
having all linear combinations of a1 , . . . , at as roots and σ0 = 1. The key equation

is:
t [ j]
si(l) = − σ j si−(l)
j , (1.15)
j=1
for l = 1, . . . , and i = t + 1, . . . , d − 1. To solve this key equation we should

solve the problem that given vectors s (1) , . . . , s () ∈ Fqd−1
m of length d − 1, find the
minimum integer t ≥ 0 such that the system of equations has a solution σ (x). A
solution can be found by solving the key equation for t = 0, 1, . . . , d − 1. The
complexity of this approach is O(d 3 ). In the following section we give an efficient
solution having complexity O(d 2 ). After the key equation is solved we calculate
vector a. After a vector a is found, the error words e(l) can be found separately for
every interleaved code.
1.3.2 Remarks
Details on these results can be found in [32]. Further details and bounds can be
found in [30]. Further, in [46], an interpolation-based unique decoder as well as a
list decoder for interleaved Gabidulin codes were presented.
1.4 Partial Unit Memory Codes
In non-coherent multi-shot RLNC, the unknown and time-variant network is used

several times. In order to create dependencies between the different shots, convolu-
tional network codes are used. Any convolutional code can be described as a Partial
Unit Memory (PUM) code. PUM codes can be constructed based on block codes.
Thus, for PUM codes over rank-metric Gabidulin codes can be used. One construc-
tion is based on the parity-check matrix of the code, and another on the generator
matrix. The constructions require a modified rank metric, the so-called sum rank
metric. For the sum rank metric, the free rank distance, the extended row rank dis-
tance and its slope were defined [40]. The construction of PUM codes based on the
parity-check matrix of Gabidulin codes achieves the upper bound on the free rank
distance.
PUM codes over rank metric can efficiently be decoded when errors, erasures and
deviations occur. The decoding complexity is cubic with the length. Moreover, we
described how lifting of these codes can be applied for error correction in RLNC
[45].
In [22], Unit Memory (UM) codes were defined as a special class of rate k/n
convolutional codes with memory m = 1 and overall constraint length ν = k. In
[21], this idea was extended to Partial Unit Memory (PUM) codes where the memory
of the convolutional code is also m = 1, but the overall constraint length is ν < k.
Known results of (P)UM codes based on block codes are e.g. Reed–Solomon
(RS) [19, 27, 50] or BCH codes [7, 8]. The use of block codes makes an algebraic
description of these convolutional codes possible. A convolutional code can be char-
acterized by its free distance and the average linear increase (slope) of the extended
row distance. These distance measures determine the error–correcting capability of
the convolutional code. In [21, 22, 27, 39] upper bounds for the free (Hamming)
distance and the slope of (P)UM codes were derived. There are constructions that
achieve the upper bound for the free (Hamming) distance, e.g. [19, 27, 50]. A con-
struction of (P)UM codes based on RS codes with optimal free distance and half the
optimal slope was given in [50].
1.4.1 Convolutional Codes
Definition 1.4.1 (Convolutional Code [40]) A rate k/n convolutional code C over
F with memory m is defined by its k × n generator matrix in polynomial form:
⎛ ⎞
g11 (D) g12 (D) . . . g1n (D)
⎜ .. ⎟ ,
G(D) = ⎝ ... ..
.
..
. . ⎠
gk1 (D) gk2 (D) . . . gkn (D)
where gi j (D) = gi(0) (1) (2) 2 (m) m ()

j + gi j D + gi j D + · · · + gi j D with gi j ∈ F, = 0, . . . , m,
for i = 1, . . . , k and j = 1, . . . , n.
We denote the constraint lengths νi , the memory m and the overall constraint
length ν of the convolutional code C according to [3] as follows.
Definition 1.4.2 (Memory and Constraint Length [40]) The constraint length for the
i–th input of a polynomial generator matrix G(D) is
def
νi = max {deg gi j (D)}.
1≤ j≤n
The memory is the maximum constraint length:
def
m = max {νi }.
1≤i≤k
The overall constraint length ν is defined as the sum of the constraint lengths νi :
def

k
ν = νi .
i=1
A codeword c(D) = c0 + c1 D + c2 D 2 + . . . of the convolutional code C is

generated by c(D) = u(D) · G(D), where u(D) = u 0 + u 1 D + u 2 D 2 + . . . is the
information.
A polynomial (n − k) × n parity–check matrix H(D) of C is defined such that for
every c(D) ∈ C, c(D) · HT (D) = 0 holds, where H(D) = [h i j (D)]1≤i≤(n−k),1≤ j≤n
with h i j (D) = h i(0) (1) (2) 2
j + hi j D + hi j D + · · · + hi j
(m H ) m H
D , where h i()
j ∈ F for
= 0, . . . , m H .
As usual, we can represent G(D) and H(D) as semi–infinite matrices G and
H over F with the submatrices G0 , G1 , . . . Gm and H0 , H1 , . . . , Hm H , where Gi ,
i = 0, . . . , m are k × n–matrices and Hi , i = 0, . . . , m H are (n − k) × n matrices. In
general, the number of submatrices in H and G is not equal, i.e., m H = m. We call
m H the dual memory of C. If both G and H are in minimal basic encoding form, the
overall number of memory elements, i.e., the overall constraint length ν is the same
in both representations [10].

c(D) and represented as ncausal infinitek sequences: c = c0 c1 c2 . . .
u(D) can be
and u = u0 u1 u2 . . . , where c j ∈ F and u j ∈ F for all j. Then, c = u · G and
c · HT = 0.
1.4.2 (Partial) Unit Memory Codes
(P)UM codes are a special class of convolutional codes with memory m = 1 [21,
22], i.e., the semi–infinite generator matrix G is given by:
⎛ ⎞
G0 G1
⎜ G0 G1 ⎟
G=⎝ ⎠, (1.16)
.. ..
. .
where G0 and G1 are k × n matrices. For an (n, k) UM code, both matrices have full
rank. For an (n, k | k1 ) PUM code, rk(G0 ) = k and rk(G1 ) = k1 < k:

G00 G10
G0 = , G1 = ,
G01 0
where G00 and G10 are k1 × n matrices and G01 is a (k − k1 ) × n–matrix. For both
cases, we have the following encoding rule:
c j = u j · G0 + u j−1 · G1 .
Note that the overall constraint length for UM codes is ν = k and for PUM codes
ν = k1 , since this is the number of symbol which influence the next block.
There are restrictions on the code rate of (P)UM codes when a certain number of
full–rank submatrices Hi should exist. This full–rank condition, rk(Hi ) = n − k for
all i = 0, . . . , m H is used in the following construction.
Lemma 1.4.1 (Rate Restriction for UM Codes [40]) An (n, k) UM code with overall
constraint length ν = k has rate
(n − k) · m H
R=
(n − k) · (m H + 1)
if the parity–check matrix H in minimal basic encoding form consists of m H + 1

full–rank submatrices Hi for m H ≥ 1.
Lemma 1.4.2 (Rate Restriction for PUM Codes [40]) Let an (n, k | k1 ) PUM code
with ν = k1 < k be given. Its rate is
k mH
R= > ,
n mH + 1
if the parity–check matrix H in minimal basic encoding form consists of m H + 1

submatrices Hi for m H ≥ 1.
If we use the parity–check matrix to construct (P)UM codes, the following theorem
guarantees that there is always a corresponding generator matrix that defines a (P)UM
code.
Theorem 1.4.1 For each semi–infinite parity–check matrix H, where the (m H + 1)

submatrices Hi are (n−k)×n matrices of rk(Hi ) = n−k, ∀i, and R ≥ m H /(m H +1),
there exists a generator matrix G such that G · HT = 0 and G defines a (P)UM code
with k1 ≥ (k − 1)/2 [40].
1.4.3 Distance Measures
Definition 1.4.3 (Sum Rank Weight [40]) Let a vector v ∈ Fn be given and let it be
decomposed into subvectors:

v = v0 v1 . . . v−1 ,
with vi ∈ Fn for all i. We define the sum rank weight wtr k (v) as the sum of the rank
norms of the subvectors:
−1

def
wtrk (v) = rk(vi ), (1.17)
i=0
for 0 ≤ ≤ ∞.
Hence, we define the sum rank distance between two sequences v(1) , v(2) of length
n by
−1

d(v(1) , v(2) ) = wt rk (v(1) − v(2) ) = dR (vi(1) , vi(2) ).
def
(1.18)
i=0
The free rank distance dfree is defined as follows.

Definition 1.4.4 (Free Rank Distance [40]) The minimum sum rank distance between
any two codewords c(1) , c(2) from a convolutional code C is the free rank distance
dfree : ∞
def (1) (2) (1) (2)
dfree =min d(c , c ) =min dR (ci , ci ) . (1.19)
c(1) =c(2) c(1) =c(2)
i=0
Generally, the error–correcting capability of convolutional codes is determined by

active distances. In the following, we define the extended row rank distance which
is an important active distance in the sum rank metric.
Let C r () denote the set of all codewords c() corresponding to paths in the minimal
code trellis which diverge from the zero state at depth j and return to the zero state
for the first time after + 1 branches at depth j + + 1. W.l.o.g., we assume j = 0
as we only consider time–invariant convolutional codes.
The extended row rank distance of order is defined as the minimum sum rank
weight of all codewords in C r ().
Definition 1.4.5 (Extended Row Rank Distance [40]) The extended row rank dis-
tance of order = 1, 2, . . . is defined as
def
dr = min wt rk (c) . (1.20)
r
c∈C ()
The minimum of the th order extended row rank distances gives the free rank
distance:
dfree = min dr .

As for Hamming metric, the extended row rank distance dr can be lower bounded by
a linear function dr ≥ max{α + β, dfree } where β ≤ dfree and α denotes the slope
(the average linear increase). The slope is an important parameter for determining
the error–correcting capability and is defined as follows (analog to [8]).
Definition 1.4.6 (Slope [40]) The average linear increase of the extended row rank
distance (slope) is defined as r
def d
α = lim . (1.21)
→∞
1.4.4 Upper Bounds
Assume, the codewords of a convolutional code are considered in Hamming metric.

Let us then denote the th order active row Hamming distance by dH,r and the free
H
Hamming distance of this convolutional code by dfree .
Theorem 1.4.2 (Connection between Distances [40]) For the free rank/Hamming
distance and the extended row rank/Hamming distance, the following holds:
dfree ≤ dfree
H
, and dr ≤ dH,r , = 1, 2, . . . .
For the rank norm rk(vi ) and the Hamming norm wt H (vi ) of a vector vi it holds that:
rk(vi ) ≤ wt H (vi ) and hence also
−1

rk(vi ) ≤ wt H (v0 . . . v−1 ).
i=0
Consequently, the upper bounds for the free distance and the slope of (P)UM
codes based on Hamming metric [27, 39] also hold for (P)UM codes based on the
sum rank metric.
Corollary 1.4.1 (Upper Bounds [40]) For an (n, k) UM code, where ν = k, the free
rank distance is upper bounded by:
dfree ≤ 2n − k + 1. (1.22)
For an (n, k | k1 ) PUM code, where ν = k1 < k, the free rank distance is upper
bounded by:
dfree ≤ n − k + ν + 1. (1.23)
For both UM and PUM codes, the average linear increase (slope) is upper boun-
ded by:
α ≤ n − k. (1.24)
1.4.5 Construction
We show now an explicit construction using normal bases.

Definition 1.4.7 ((P)UM Code Based on Gabidulin Codes [40]) For some m H ≥ 1,
let a rate R = k/n = c · m H /(c · (m H + 1)) UM code or a rate R = k/n >
m H /(m H + 1) PUM code over F be defined by its semi–infinite parity–check matrix
H. Let each submatrix Hi , i = 0, . . . , m H be the parity–check matrix of an (n, k)
Gabidulin code G (i) :

Hi = Vn−k (h(i) ) = Vn−k h 1(i) h 2(i) . . . h n(i) ,
for all i = 0, . . . , m H . Additionally, let

⎛ ⎞
H0
def ⎜ H1 ⎟
H(c) =⎜ ⎟ (c)
⎝ . . . ⎠ define G , (1.25)
Hm H
def
H(r (i)) = Hi Hi−1 . . . H0 define G (ri ) , (1.26)
for all i = 1, . . . , m H . G (c) is an (n (c) , k (c) ) Gabidulin code and G (ri ) is an (n (ri ) , k (ri ) )
Gabidulin code with
n (c) = n, k (c) = n − (m H + 1)(n − k),

n (ri ) = (i + 1)n, k (ri ) = i · n + k, i = 1, . . . , m H .
Hence, not only each submatrix has to define a Gabidulin code, but also the rows
and columns of submatrices of H.
Now, we give an explicit construction that fulfills the requirements of Defini-
tion 1.4.7. To ensure that (1.25) is fulfilledand H(c) defines a Gabidulin code, H1
has to be the continuation of H0 , i.e., h(1) = (h (0)
1 )
[n−k]
, (h (0)
2 )
[n−k]
, . . . , (h (0)
n )
[n−k]
.
(2) (1)
Also, h has to be the continuation of h and so on. Hence,

H(c) = V(m H +1)(n−k) h(0)

= V(m H +1)(n−k) h (0) (0) (0) .
1 h2 . . . hn (1.27)
In order to fulfill (1.26), we have to ensure that all elements from F in the set
H = {h (0) (0) (1) (1) (m H )

, . . . , h n(m H ) }
def
1 , . . . , hn , h1 , . . . , hn , . . . , h1
= {h (0) (0) (0) [n−k]

1 , . . . , h n , (h 1 ) , . . . , (h (0)
n )
[n−k]
, (1.28)
. . . , (h (0)
1 )
[m H (n−k)]
, . . . , (h (0)
n )
[m H (n−k)]
}
with |H| = (m H + 1) · n are linearly independent over Fq .

To obtain an explicit construction of such a (P)UM code, we can use a normal
basis. A basis B = {b0 , b1 , . . . , bs−1 } of F over Fq is a normal basis if bi = b[i] for
all i and b ∈ F is called a normal element. There is a normal basis for any finite
extension field F [23]. For our construction, we use a normal element b to define

h(0) = b[0] . . . b[n−k−1] b[(m H +1)(n−k)] . . . b[(m H +2)(n−k)−1]

b[2(m H +1)(n−k)] . . . b[2(m H +2)(n−k)−1] . . . . (1.29)
This h(0) is used to define H(c) (1.27) and hence also H is defined (1.25).
To make sure that also (1.26), (1.28) are fulfilled, we require a certain minimal
field size. If (n − k) divides n, then h(0) can be divided into subvectors, each of length
(n − k) (1.29) and the field size has to fulfill s ≥ (m H + 1) · n to ensure that all
elements in H are linearly independent (1.28) and that (1.26) is fulfilled. If (n − k)
does not divide n, the last subvector in h(0) is shorter than n − k. Equations (1.28)
and (1.26) can be
⎛ ⎞
a [0] a [1] a [4] a [5] a [8] a [9]
⎜ a [1] a [2] a [5] a [6] a [9] a [10] ⎟
⎛ ⎞ ⎜ ⎟
H0 ⎜ ⎟
⎜ ⎟
⎜ H1 H0 ⎟ ⎜ a [2] a [3] a [6] a [7] a [10] a [11] a [0] a [1] a [4] a [5] a [8] a [9] ⎟
⎜ ⎟ ⎜ ⎟
H=⎜ . ⎟ = ⎜ a [3] a [4] a [7] a [8] a [11] a [12] a [1] a [2] a [5] a [6] a [9] a [10] ⎟.
⎜
⎝ H1 . . ⎟ ⎜
⎠ ⎜
⎟
⎟
.. ⎜ ⎟
. ⎜ a [2] a [3] a [6] a [7] a [10] a [11] ...⎟
⎜ ⎟
⎝ a [3] a [4] a [7] a [8] a [11] a [12] ...⎠
(1.30)
Another random document with
no related content on Scribd:
as was the case years ago, an unhealthy excitement in the minds of
220
the people.”
But, facts being more convincing than official denials, the exodus
grew more alarming, because the forces to which it owed its origin
continued in operation. The “Jewish Colonization Association” now
came to the aid of the indigent exiles, and endeavoured to save
them from additional suffering by preventing those who were not
provided with the necessary passage money, or were not physically
221
fit, from leaving their homes. These wise measures restrained to
a certain extent indiscriminate expatriation, but, as might have been
foreseen, failed to check it entirely. The exodus continued, and the
outcry against Roumania spread, for now the countries into which
the undesirable current flowed were compelled by self-interest to do
what they had hitherto vainly attempted to effect from a sense of
philanthropy.
America, the favourite haven of refuge for the fortune-seeker of
every colour and clime, undertook the task of spokesman. The late
Mr. Hay, Secretary of State, in September, 1902, through the
representatives of the United States in the countries which took part
in the Congress of Berlin, reminded the Governments of those
countries of Art. 44 of the Treaty signed by them in 1878, urging
them to bring home to Roumania her flagrant and persistent failure
to fulfil the conditions on which she had obtained her independence.
After a handsome tribute to the intellectual and moral qualities of the
Jew, based on history and experience, the American Minister
protested, on behalf of his country, against “the treatment to which
the Jews of Roumania are subjected, not alone because it has
unimpeachable ground to remonstrate against resultant injury to
itself, but in the name of humanity.” He concluded with a vigorous
appeal to “the principles of International Law and eternal justice,”
and with an offer to lend the moral support of the United States to
222
any effort made to enforce respect for the Treaty of Berlin.
This powerful impeachment, coming as it did from a distant party
in no way connected with the affairs of Continental Europe, may
have caused heart-searchings in nearer and more immediately
concerned countries; but it failed to awaken those countries to a
proper sense of their interests, not to say duties. The only quarter in
which America’s appeal to humanity found an echo was England. A
number of representative men, such as the late Archbishop of
Canterbury, the present Bishop of London, Lord Kelvin, the
Marquess of Ripon, the late Mr. Lecky, Sir Charles Dilke, the Master
of Balliol, and others, publicly expressed their profound sympathy
with the victims of persecution. Mr. Chamberlain also seized the
opportunity of declaring that, as history proves, the Jews, “while
preserving with extraordinary tenacity their national characteristics
and the tenets of their religion, have been amongst the most loyal
subjects of the states in which they have found a home, and the
impolicy of persecution in such a case is almost greater than its
223
cruelty.” Other Englishmen also joined in the denunciation of
Roumania not so much from pity for the victims of oppression as
from fear lest, unless the Roumanian Government was compelled to
change its policy, England should have to face another inroad of
“undesirable” Jewish immigrants.
In like manner, the only Government which volunteered to
second Mr. Hay’s Note was the British, and on the common basis of
these two representations, the signatory Powers of the Treaty of
Berlin “exchanged views.” The results of this exchange can be
summed up only too easily. The historian of the future will probably
derive therefrom some interesting lessons regarding European
politics and ethics in the beginning of the twentieth century. They are
as follows:
Germany, under whose presidency the stipulation concerning the
Jews of Roumania was framed, did not choose to consider herself
called upon to insist on the execution of that stipulation. The Liberal
section of the German press received the American Note with
sincere, but ineffectual, appreciation; while of the Conservative
majority some pronounced it naïve, and others affected to regard it
as an attempt on America’s part to interfere in European affairs, or
even as an electioneering trick having for its sole object to enhance
President Roosevelt’s political prestige! The German Government,
though more courteous than the German press, proved equally cold.
As we have already seen, that Government was the last to join in the
efforts to improve the lot of the Roumanian Jews and the first to
declare itself satisfied with the deceptive revision of Article 7 of the
Roumanian Constitution. This attitude, when considered in
conjunction with the fact that a Hohenzollern reigns in Roumania,
and with that kingdom’s place in the present political combinations of
the Continent, enables us to understand, if not to applaud,
Germany’s reception of Mr. Hay’s Note.
Austria-Hungary, whose proximity to Roumania pointed her out
as the Power primarily concerned, and entitled to act, declined to
take any steps singly or collectively. The self-restraint of Austria, like
that of Germany, and even in a greater degree, was dictated by
political considerations, Roumania being practically the only State in
the Balkans, where the influence of Austria-Hungary and of the Triple
Alliance still counts for something. Besides, the Vienna Cabinet
could not decently join in advocating Jewish emancipation, for it was
Austria which in May, 1887, concluded with Roumania a treaty
whereby some seventy thousand Jewish residents in the latter
kingdom—who, according to a practice common in Mohammedan
countries, had enjoyed Austrian protection while Roumania was
under Ottoman rule—were deprived of the status of Austrian
subjects, without receiving any other status in exchange.
Italy was deterred from lending her support to the American Note
by Roumania’s relations with the Triple Alliance and also by the
vogue which the “Roman” idea obtains in the land which the
Roumanians are pleased to regard as “the cradle of their race.”
Russia, whose treatment of her own Jewish subjects would have
made an appeal to “humanity and eternal justice” on behalf of the
Jews in another country a sad mockery, decorously refrained from
supporting the American Note. It is true that the Russian press
imitated the Teutonic in scoffing at America’s action as a pretext for
gaining admission to the counsels of the European Areopagus, and
in condemning it as an impertinence! But the Czar’s Government,
with better taste, extricated itself from an awkward position by basing
its refusal on the ground that the grievances set forth in Mr. Hay’s
despatch were so old that it was hardly worth while troubling about
them. In the opinion of the Russian Ministers, the Jews must by now
be thoroughly accustomed to starvation.
France, with all the good intentions in the world, could do nothing
without Russia’s consent and, therefore, contented herself with the
expression of a modest hope that the Roumanian Government might
of their own accord decide to fulfil their obligations, seeing that the
real sufferer is Roumania itself, and with pointing to the lack of
224
means of enforcing such fulfilment.
In brief, the European Powers considered that they did their duty
by expressing their platonic concurrence with that part of the
American Note which referred to the obligations of humanity and
civilisation generally. But to the more definite appeal to the Treaty of
Berlin they refused to pay any attention whatsoever. Nor can we
wonder at their refusal. The appeal was not a very happy one; for
every party to that contract has conscientiously broken it in turn.
Russia, in defiance of its provisions, has fortified Batoum; Turkey has
not even attempted to carry out the reforms in the European
Provinces of the Empire, ordained by the Treaty; Great Britain has
done nothing for the Armenians. Why then should poor Roumania
alone be called upon to carry out her share of an agreement, already
disregarded with impunity by everyone else concerned?
Such a retort would, of course, have been too candid and too
rational for diplomacy. Instead, the Roumanian Government had
again recourse to the more correct, if somewhat hackneyed,
expedient of an official contradiction of the truth. The Roumanian
Minister in London declared that “the idea that any persecution
existed was absolutely erroneous.” The Jews were foreigners, and
“the disabilities imposed upon foreigners were absolutely necessary
for the protection of his countrymen, who had bought their
independence with the sword, and had a right to manage their
225
economic affairs according to their requirements, etc., etc.” What
the Roumanian conception of such a right is has been very
eloquently explained by Roumania’s accomplished Queen. After
having drawn a pitiful and, although exaggerated, in the main faithful
picture of Roumania’s economic misery, Her Majesty declares that,
under such conditions, the civilised world ought not “to require her to
harbour and support others, when she herself stands in dire need of
assistance.” Those “others” are “foreigners,” that is, Roumanian
Jews; their exodus is represented as the voluntary emigration of “a
foreign population” due to the instinct which prompts a rat to quit a
sinking ship, and their departure is welcome, because they, being
traders, drain the country of its wealth. This interesting economic
doctrine is expounded by Her Majesty as follows: “It is a fact that no
money has ever been introduced into Roumania through any one in
trade. Any that such a man may possess goes abroad, first to
purchase his stock and outfit, and later for supplies to carry on his
business, even such articles as buttons and the commonest kinds of
braids not being manufactured here except on the very smallest
226
scale.” Here again the Jewish apologist is more convincing than
his Roumanian accuser. Admitting that, on the whole, the Queen’s
statements are correct, he asks: “But why is it so? For the reason
that the ruling class prohibits ‘foreigners’ to acquire lands in the
country, and by means of this and other laws keeps foreign capital
227
from coming in.”
Protests pass away, grievances remain. The well-meant action of
Mr. Hay and Lord Lansdowne, far from bettering, really aggravated
the condition of the people on whose behalf it was taken. The
Roumanian politicians, with characteristic astuteness, perceived that
the immediate cause of the complaint was the emigration of the
Jews to the United States, England and Canada, and, naturally
enough, arrived at the conclusion that the one thing needful was to
remove the ground of complaint by stopping emigration. A
telegraphic order was sent to all the local authorities, forbidding the
issue of passports to the Jews. Those who had already reached the
frontier were forcibly turned back, and hundreds of others, who had
sold all they possessed in order to raise the funds necessary for the
228
journey, were compelled to return home and perish. Thus an act
intended as a blessing proved an unmitigated curse, and modern
Roumania by this new measure has outstripped even mediaeval
Spain in cruelty. For the Spanish sovereigns, blinded by religious
bigotry, had yet given to the Jews the alternatives of conversion or
exile. Their Roumanian imitators, infatuated by racial fanaticism, will
not baptize the Jews, nor dare they banish them; but, like Pharaoh of
old, they virtually bid them stay and be slaves.
CHAPTER XXIII
ANTI-SEMITISM
We have followed the fortunes of the Jewish people from the

moment of its first contact with the nations of the West to the last
quarter of the nineteenth century. We have seen that this contact
was from the beginning marked by mutual antipathy, enfeebled at
times, invigorated at others, always present. Some Jewish writers
have endeavoured to show that the hatred of the Gentile towards the
Jew in the Middle Ages was an artificial creation due entirely to the
efforts of the Catholic Church; that it flowed from above, and that the
masses of Christendom, when not incited by the classes, were most
amicably disposed towards Israel. This view is hardly tenable. It is
inconceivable that the Church, or any other authority, could have
succeeded so well in kindling the conflagrations which we have
witnessed, if the fuel were not ready to be kindled. It is also a view
contrary to the recorded facts. We have seen in the earlier Middle
Ages popular prejudice spontaneously manifesting itself in the insults
and injuries which were heaped upon the Jews, and restrained with
difficulty by the princes and prelates of Europe. In the time of the
Crusades also it was not St. Bernard who fanned the fury of the mob
against the Jews of the Rhine, but an obscure monk. The
exhortations of the saint were disregarded; but the harangues of the
fanatic found an eager audience, simply because they were in
accord with popular feeling. During the same period bishops and
burgomasters strove to save the victims, in vain.
Again, the persecution of the Spanish Jews in the fifteenth
century would never have attained the dimensions which it did attain,
were it not for the deep-rooted animosity which the bulk of the
Spanish people nourished against them. Castile was then the home
of chivalry and charity. The pretensions of the Pope to interfere in the
affairs of the kingdom had met with scornful opposition on the part of
the Castilian nobles. Three centuries before an Aragonese monarch
had given away his life in defence of the persecuted heretics of
Provence. Less than two centuries before Aragon was one of the few
countries that refused to comply with the joint request of Philip the
Fair of France and Pope Clement V. to persecute the Knights
Templars. At the time when the Inquisition was established in Spain
both Castile and Aragon were hailing the revival of culture. Under
Ferdinand and Isabella, as well as in the subsequent reigns, the
Castilians and the Aragonese vigorously resisted an institution so
contrary to the principles of freedom dear to them. Nor was in Spain
the danger of dissension sufficiently great to justify recourse to so
terrible an instrument of concord. The Spaniards less than any other
people had reason to sacrifice liberty of conscience for the sake of
political conquest. It is, therefore, highly improbable that the Holy
Office would ever have gained a firm footing in Spain, but for the fact
that its way was paved by the popular prejudice against the Jews
and the Moors, and its success assured by the persecution of those
races. Though the Spaniards hated the Inquisition bitterly, they hated
the Semites more bitterly still; and of the two the Jew more bitterly
than the Moor.
We have also seen that neither the Renaissance nor the
Reformation, both movements directly or indirectly hostile to the
Church, brought any amelioration to the lot of the Jew. In every
country Jew-hatred existed as the product of other than
ecclesiastical influences. Here and there, under exceptionally
favourable conditions, the Jews may have been tolerated; they were
not loved. This negative attitude was liable to be at any moment
converted into active hostility. All that the Church did was to turn the
feeling to account, to intensify and to sanctify it. Lastly, we have seen
that the emancipation of the Jews did not come about until the end of
the eighteenth and the middle of the nineteenth century—a period no
longer of protest against the Church, but one of rebellion against all
the prejudices of all the ages. It was not until the gospel of humanity,
in its broadest sense, was accepted that the secular clamour against
the Jewish portion of the human race was silenced; and even then
not without difficulty. But, though the plant of anti-Judaism was cut at
the root, the root remained, and it was destined in our own day to put
forth a new shoot.
Writers have expended much ingenuity in defining the origin and
the nature of modern anti-Semitism. Some regard it as a
resuscitation of mediaeval religious bigotry; others as the latest
manifestation of the old struggle between Europe and Asia; a third
school, rejecting both those theories, interprets it as a purely political
question arising from the social and economic conditions created by
the emancipation of the Jews; while a fourth sect have attempted to
show that the modern revival is “the fruit of a great ethnographical
and political error.” Those who see in anti-Semitism nothing but a
revival of mediaeval religious rancour ignore the conflict between
Jew and Gentile before the rise of the Mediaeval Church, or even
before the rise of Christianity. Those who explain it as a purely racial
struggle forget the Crusades and the Inquisition and the superstitious
horror of usury. Those who interpret it simply as a question of
modern European politics disregard both those periods of history.
Finally, whatever may be said of crude ethnographical theories and
of nebulous nationalist creeds, it would be doing them too much
honour to suppose that they are the real causes of anti-Semitism.
Men do not slaughter their fellow-men for the mere sake of an
abstract hypothesis, though priests may. All these things do nothing
but give a name and a watchword to a movement born of far less
ethereal parents. In our day the political activity which has used anti-
Semitism as an instrument has only done what clerical activity had
done in the past. It has availed itself of a force not of its own
creation. The fact is that every human action is the result of manifold
motives. The complexity of the motives is not diminished by the
multitude of the actors. There is a strong temptation to simplify
matters by singling out one of those motives and ignoring the rest.
But, though truth is always simple, simplicity need not always be
true. There may be new things under the sun. Anti-Semitism,
however, is not one of them. Its roots lie deep in the past.
Viewed, then, in the light of two thousand years’ recorded
experience, modern anti-Semitism appears to be neither religious,
nor racial, nor economical in its origin and character. It is all three,
and something more. We find in it all the motives which led to the
persecution of the Jews in the past. In antiquity the struggle was
chiefly due to racial antagonism, in the Middle Ages chiefly to
religious antagonism, in the nineteenth century we might expect it to
assume chiefly a nationalist garb. But, as in antiquity religious
antipathy was blended with racial hatred, as in the Middle Ages
economic rivalry accentuated religious bigotry, so in our time
religious, racial, and economic reasons have contributed to the
movement in various degrees according to the peculiar conditions,
material and moral, prevailing in each country where anti-Semitism
has found an echo. If it were possible to unite all these causes in one
general principle, it would be this: every age has its own fashionable
cult, which for the time being overshadows all other cults, gives a
name to the age, explains its achievements, and extenuates its
crimes. Every age has found in the Jew an uncompromising
dissenter and a sacrificial victim. The cult par excellence of the
nineteenth century is Nationalism.
What is this dreadful Nationalism? It is a reversion to a primitive
type of patriotism—the narrow feeling which makes men regard all
those who live in the same place, or who speak the same language,
or who are supposed to be descended from a common ancestor, as
brethren; all others as foreigners and potential foes. This feeling in
its crudest form is purely a family-feeling, in the worst sense of the
term. It grows into a larger allegiance to the tribe, then to the race,
and that in its turn develops into the broad patriotism which
manifests itself now as Imperialism, now as Catholicism.
There is yet a third form of patriotism—the purest and noblest of
all: loyalty to common intellectual ideals. The Greeks attained to this
lofty conception, and an Athenian orator, in enumerating his
country’s claims to the admiration of mankind, dwells with just pride
on this product of its civilisation. Athens, he says, “has made the
name of the Hellenes to be no longer a name of race, but one of
mind, so that Hellenes should be called those who share in our
229
culture rather than in our nature.” Isocrates in making this
statement, however, gave utterance to a dream of his own rather
than to a feeling common among his countrymen. The Macedonian
Empire strove to convert that philosophical dream into a political fact.
Alexander and his successors studded Asia with Greek theatres,
Greek schools, Greek gymnasia, and the East was covered with a
veneer of pseudo-Hellenic civilisation. But their success was only
partial, superficial and ephemeral. The intellectual unity could not go
deep and therefore did not last long. The barriers—social, religious
and racial—which separated the Hellene from the Barbarian proved
insuperable; and the Isocratean ideal of a nationality based on
community of intellectual aims remained an ideal. Hellenism
demanded a degree of mental development to which mankind has
never yet attained. Hence its failure as a political bond. This was not
the case with Imperialism and Catholicism. They both appealed to
more elementary and therefore less rare qualities in man. Hence
their success. Rome achieved more than Greece because she
aimed at less.
The Roman Empire represented the first, the Roman Church the
second variety of this broad patriotism. Civis Romanus was a title
which united in a common allegiance the Italian and the Greek, the
Jew and the Egyptian, the Spaniard, the Briton and the Gaul.
Catholic Rome inherited the imperial feeling of Pagan Rome, but
dressed it in a religious form. The dictatorship of the Caesars was
divided between the Christian Emperor and the Pope: the former
inheriting their political power, the latter the spiritual and moral.
Charlemagne wielded the authority of an Imperator Romanus, his
papal contemporary that of a Pontifex Maximus. Then came the
decay and fall of the Carlovingian fabric; and, gradually, the Papacy
built up a spiritual empire with the débris of the secular. All Catholics
were subjects of that Empire. In the Middle Ages Europe presented a
picture of wonderful uniformity in sentiments, ideals, customs,
political and social institutions. All countries, like so many coins
issued from one mint, seemed to be cast in the same mould,
stamped with the same effigy and adorned with the same legend.
National consciousness was in the Middle Ages practically non-
existent, or, if it did exist, in the later centuries, it was obscured by
the religious sentiment. As in modern Islam we find Arabs, Persians,
Indians, Malays, Chinese, Syrians, Egyptians, Berbers, Moors,
Turks, Albanians—nations differing widely in origin and language—
united by the ties of a common creed, so in mediaeval Christendom
we find English, Scotch, French, Italian, German and Spanish
knights all forming one vast brotherhood. The reader of Froissart
cannot fail to notice this community of feeling and the marvellous
ease with which gentlemen from all those nations made themselves
at home in one another’s countries. The chronicler himself, in his
style and mental attitude, supplies a striking example of this
cosmopolitanism. By the mediaeval Christian, as by the modern
Mohammedan, the human race was divided into two halves: true
believers and others. The universal acceptance of Latin as the
medium of communication was another token and bond of
brotherhood among the Christians of mediaeval Europe, as the use
of Arabic, as a sacred tongue, is a token and a bond of brotherhood
among the Mohammedans of the present day.
This feeling of international patriotism, which found its highest
development and expression in the Crusades, began to fade as soon
as Catholic faith began to decay. Disintegration followed both in the
Church and in the State. Loyalty to one ideal and to one authority
was gradually superseded by local and later by racial patriotism.
Various political units succeeded to the Unity of mediaeval Europe,
the vernaculars ousted the Latin language from its position as the
one vehicle of thought, and the old cosmopolitan universities of Paris
and Bologna were replaced by national institutions. Since the
fifteenth century nationalism has been growing steadily, but in the
eighteenth its growth was to some extent checked by
humanitarianism. The great thinkers of that age extolled the freedom
and the perfection of the individual as the highest aim of culture,
describing exclusive attachment to one’s country and race as a
characteristic of a comparatively barbarous state of society: a
remnant of aboriginal ancestor-worship. Nationalism, accordingly, did
not reach its adolescence until the nineteenth century. Then the zeal
for peace was eclipsed by the splendour of the French exploits in
war, and the doctrine of universal freedom was forgotten in
Napoleon’s efforts at universal dominion. These efforts aroused in
every country which Napoleon attacked a passionate protest which
resulted in successful revolt. But the triumph was won at a
tremendous cost. Each nation in proportion to its sense of what was
due to itself was oblivious of what was due to others. The principles
of the brotherhood of men and of universal toleration were denied,
the narrow jealousies of race which the philosophers of the
preceding century had driven from the realm of culture were re-
installed, and Nationalism—arrogant, intemperate, and intolerant—
arose on the ruins of Humanitarianism. This evolution, or revolution,
has added a new element in social troubles, and has brought into
being a new set of ideas.
For the last hundred years ethnographical theory has dominated
the civilised world and its destinies as theological dogma had done
during the Middle Ages. Consciously or not, the idea of race directs
the policy of nations, inspires their poetry, and tinges their philosophy
with the same prejudice as religion did formerly. Aryan and non-
Aryan have become terms conveying all but the odious connotation
of Christian and infidel; and in place of the spiritual we have adopted
a scientific mythology. The fiction of our Aryan origin has flattered us
into the benevolent belief of our mental superiority over the Mongol,
and of our moral superiority over the Semite. To dispute this tenet is
to commit sacrilege. But even within the bosom of this imaginary
Aryan fold there are schisms: so-called Celtic, Germanic, Latin,
Anglo-Saxon, and Slavonic sects, divided against one another by the
phantom barriers of ethnographical speculation as frantically as in
older days Christendom was divided by the metaphysical figments of
Arian, Manichaean, Nestorian, and what not. In the name of race are
now done as many great deeds and as many great follies are
committed as were once in the name of God. The worship of race
has, as the worship of the Cross had done before, given birth to new
Crusades which have equalled the old in the degree to which they
have disturbed the peace and agitated the minds of men, and in the
violence of the passions which they have excited. Nationalism more
than any other cause has helped to bring discredit upon the
principles of liberty, equality, and fraternity—to prove the eighteenth
century dream of world-wide peace a glorious impossibility—and to
show the enormous chasm which still gapes between the aspirations
of a few thinkers and the instincts of the masses.
Though common to all European countries, the creed of the age
found articulate exposition first in Germany, and gave rise to various
academic doctrines which attempted to account for the genesis and
evolution of Nationalism in scientific or pseudo-scientific terms. But
names do not alter facts. Ethnographical speculations are in this
case mainly interesting as having supplied a plausible explanation
for the rise of anti-Semitism. Those who are able to see through new
guises, and to detect what old things they conceal, know that anti-
Semitism is little more than a new Protean manifestation of Jew-
hatred. Divested of its academic paraphernalia, the movement is
revealed in all its venerable vulgarity—a hoary-headed abomination
long since excommunicated by the conscience of civilised mankind.
This reactionary movement began in Eastern Germany and
230
Austria. In those countries the Jews are very numerous, very
wealthy, and very influential. Both countries are famous as hot-beds
of racial fanaticism. In Germany Nationalism was begotten of the
independence secured by the Thirty Years’ War in the seventeenth
century, was nursed by the patriotic preachers and poets of the
eighteenth, was invigorated by the wars for emancipation from
Napoleon’s rule, and was educated by Hegel and his disciples. The
Jews in Germany, as elsewhere, are the one element which declines
to be fused in the nationalist crucible. Their international connections
help them to overstep the barriers of country. Their own racial
consciousness, fostered by the same writers, is at least as intense
as that of the Germans; but it does not coincide with any
geographical entity. They are, therefore, regarded as a cosmopolitan
tribe—“everywhere and nowhere at home.” They are distinct not only
as a race, but as a sect, and as a class. Accordingly, the reaction
against tolerance includes in its ranks clerics and Christian
Socialists, aristocrats, as well as Nationalists, that is, the enemies of
dissent and the enemies of wealth, as well as the enemies of the
alien and the enemies of the upstart. And the term “Jew” is used in a
religious or a racial sense according to the speaker. In both Germany
and Austria we saw that the philosophical gospel of social liberty was
very slowly applied to practical politics, and that, even when it had
been accepted, it was subject to reactions. When Jewish
manumission was finally accomplished, the Jews by their genius
filled a much larger place in the sphere of national life than was
deemed proportional to their numbers. And this undue
preponderance, rendered all the easier by the superior cohesion of
the Jewish over the German social system, was further accentuated
by specialisation. The Jews, whose training in Europe for centuries,
owing partly to their own racial instincts and Rabbinical teaching, but
chiefly to the conditions imposed upon them from outside, had been
of a peculiar kind, showed these peculiarities by their choice of fields
of activity. They abstained from the productive and concentrated
their efforts to the intellectual, financial, and distributive industries of
the countries of which they became enfranchised citizens. Jews
flooded the Universities, the Academies, the Medical Profession, the
Civil Service, and the Bar. Many of the judges, and nearly one-half of
the practising lawyers of Germany, are said to be Jews. Jews came
forth as authors, journalists, and artists. Above all, Jews, thanks to
the hereditary faculty for accumulation fostered in them during the
long period when money-dealing was the one pursuit open to them,
asserted themselves as financiers. It is impossible to move
anywhere in Berlin or Vienna without seeing the name of Israel
written in great letters of gold not only over the shops, but over the
whole face of German life. Success awakened jealousy, and
economic distress—due to entirely different causes—stimulated it.
What if the competition was fair? What if the Jews were
distinguished by their peaceful and patriotic attitude? What if they
supplied the least proportion of criminals and paupers? What if
German freedom had been bought partially with Jewish blood, and
German unity achieved by the help of Jewish brains and Jewish
money?
The landed gentry, richer in ancestors than in money or
intelligence, had every reason to envy the Jew’s wealth, and much
reason to dislike the Jew’s ostentatious display of it. They could not
respect in the Jew a gifted arrivé. They saw in him a vulgar parvenu
—one who by his “subversive Mephistophelian endowment, brains,”
demolishes the fences of creed and caste, and invades the highest
and most exclusive circles, thus acting as a solvent in society. If he is
wise, the proud nobleman of narrow circumstances makes his pride
compensate for his poverty, and magnanimously despises the
luxuries which he cannot procure. If, as more often happens, he is
foolish, he enters into a rivalry of vanity with the upstart, and the
result is a mortgaged estate—mortgaged most likely to his rival. In
either case, he can have little love for the opulent and clever
interloper. The animosity of the aristocracy is shared by the middle
classes, and for analogous reasons. The German professional man,
and more especially his wife, resents his Jewish colleague’s
comparative luxury as a personal affront. The excessive power of
money in modern society, and the consequent diminution of the
respect once paid to blood or learning, naturally enable the Jewish
banker to succeed where the poor baron fails; and the Jewish
professor or doctor, though many of these latter are poor enough, to
outshine his Christian competitor. This excessive power of money is
due to causes far deeper than the enfranchisement of the Jews. It is
the normal result of Germany’s modern development. The influence
of the nobles depended largely on their domains of land; and when
industries arose to compete with agriculture, the importance of land
necessarily declined. At the same time, industry and commerce
began, with Germany’s expansion, to divert more and more the
attention of the intelligent from the path of academic distinction—
once the only path to honour open to the ambitious burgher—into
that of material prosperity. Chrematistic enterprise has introduced a
new social standard, and an aristocracy of wealth has come to
supplant the old aristocracies of birth and erudition. This social
revolution, through which every country in the world has passed and
has to pass, was unhesitatingly ascribed to the Jew, who was thus
accused of having created the conditions, which in reality he had
only exploited.
If from the aristocratic and the cultured classes we turn to the
rural population, we find similar causes yielding similar results. In the
German country districts it is objected to the Jews not cultivating the
land themselves, but lying in wait for the failing farmer:
“Everywhere,” says an authority, “the peasant proprietor hated the
Jew,” and he proceeds to sketch the peasant tragedy of which that
hatred was the consequence. The land had to be mortgaged to pay
family claims; the owner had recourse to the ubiquitous and
importunate money-lender; the money-lender, whose business it is to
trade upon the necessity of the borrower, took advantage of the
latter’s distress, and extorted as much as he could. “The Jew grew
fat as the Gentile got lean. A few bad harvests, cattle-plague, or
potato-disease, and the wretched peasant, clinging with the
unreasonable frantic love of a faithful animal to its habitat, had, in
dumb agony, to see his farm sold up, his stock disposed of, and the
acres he had toiled early and late to redeem, and watered by the
sweat of his stubborn brow, knocked down by the Jewish interloper
231
to the highest bidder.” In the Austrian country districts it is urged
that the presence of the Jew is synonymous with misery; his
absence with comparative prosperity. In Hungary, the late M. Elisée
Reclus—the famous author of the Nouvelle Géographie Universelle
—informs us, “The rich magnate goes bankrupt, and it is almost
always a Jew who acquires the encumbered property,” and another
witness adds: “The Jew is no less active in profiting by the vices and
necessities of the peasant than by those of the noble.” In Galicia,
especially, we are told that the land is rapidly passing into the hands
of the Jews, and that many a former proprietor is now reduced to
work as a day-labourer in his own farm for the benefit of a Jewish
master. All this is an absurdly exaggerated version of facts in
themselves sad enough. The Jews as a whole are by no means a
wealthy community, and the gainers by the supposed exploitation
are the few, not the many. And if, as is the case, the condition of
affairs in agricultural states is bad, who is to blame? Wherever there
is agrarian depression there are sure to be money-lenders enough
and Shylocks too many. It does not appear that Christian money-
lenders have ever been more tender-hearted than their Jewish
confrères. Why then set down to the Jew, as a Jew, what is the
common and inevitable attribute of his profession? The ruin of the
borrower does not justify the slaughter of the lender. Philanthropists
would be better employed if, instead of bewailing in mournful
diatribes the woes of the bankrupt peasant and inveighing against
the cruelty of his oppressor, combined to establish agricultural banks
where the farmer could obtain money at less exorbitant interest. This
measure, and measures like this, not slaughter and senile
lamentation, would be a remedy consonant both with the nature of
the evil and with the dictates of civilisation and justice. Until
something of the sort is done, it is worse than futile to demand that
dealers in money, any more than dealers in corn, cotton, or cheese,
should work from altruistic motives. But nothing rational is ever
attempted. Instead, everywhere the nobles ruined by their own
improvidence and extravagance, the peasants by their rustic
incompetence, and both by the exactions of a wasting militarism,
complain of the extortion of the Jewish usurers. It was inevitable that
the old-world monster of Jew-hatred, never really dead, should have
raised its hoary head again. All the elements of an anti-Jewish
movement were present. The only thing that lacked was opportunity.
The deficiency was not long in being supplied.
The Franco-German war and the achievement of German unity
fanned the flame of patriotism. As in the time of Napoleon the First,
so in that of Napoleon III., a great national danger created a strong
fellow-feeling between the different members of the German race; a
great national triumph stirred up an enthusiasm for the Empire which
was indulged in at the cost of individual liberty. Despotism throve on
the exuberance of nationalism. The Germans were led back from the
constitutional and democratic ideals of 1848 to an ultra-monarchic
servility which made it possible for the present Kaiser’s grandfather a
few years after, prompted by Bismarck, to assert openly the
ridiculous old claim to divine right. Thus the ground was prepared for
any anti-alien and anti-liberal agitation. Other causes came to
accelerate the movement. The war had involved enormous
pecuniary and personal sacrifices. The extraordinary success,
instead of satisfying, stimulated German ambition. It aroused an
extravagant financial optimism and self-confidence. Germany,
intoxicated with military victory, was still thirsting for aggrandisement
of a different kind. Economy was cast to the winds, and a fever of
wild speculation seized on all classes of the community. Companies
were floated, and swallowed up the superfluous capital of the great
as well as the savings of the humble. Sanguine expectation was the
temper of the day. Berlin would vie with Paris in elegance and with
London in suburban comfort, and every one of its citizens would be a
millionaire!
Then came the terrible crash. The bubble burst, and the
magnificent day-dreams were dispelled by misery. A succession of
bad harvests, and the rapid increase in American corn competition,
by impoverishing the agricultural class, added to the general
depression. The disillusioned public wanted a victim whereupon to
vent its wrath. Those who promoted the companies had to suffer for
the folly of those who were ruined by their failure. A great many of
the former, by selling out at the right moment, rose to affluence. The
discontented public, naturally enough, noticing these large fortunes
in the midst of the general wreck, jumped to the conclusion that the
few had enriched themselves by robbing the many. “Exposures”
followed, and among the implicated financiers there were found
many Jews. It was then in order to fill Jewish pockets that the heroes
of Germany had bled on the battlefield, and the burghers of
Germany had been bled at home! The nationalist ideal of Germany
for the Germans, then, was to lead to a Germany for the Israelites!
All those trials had been endured and all those triumphs achieved in
order to deliver up the Fatherland to an alien and infidel race—a race
with which neither the intellect nor the heart of Germany has any
affinity or sympathy! This was the cry of anguish that succeeded to
the paeans of self-glorification, and those nationalists who uttered
these sentiments forgot that their very nationalism had been largely
created and fostered by Jewish thinkers. They also forgot that it was
a Jewish statesman, Lasker, who, at the cost of all personal and
party interests and of his popularity, had alone had the courage to
expose in the Prussian Chamber the evils of extravagant
speculation, in 1873, and to urge both the public and the
Government to turn back, while there was yet time, from the road to
ruin which they pursued. But it has been well said: “Who would think
of gratitude when a scapegoat is required?”
A tongue was given to the popular indignation in a pamphlet by
an obscure German journalist, Wilhelm Marr by name, who seized
the opportunity of attaining to fame and fortune by a plentiful effusion
of his anti-Jewish venom. The work anathematized the Jews not only
as blood-sucking leeches, but as enemies of the Germanic race, and
as forming a distinct and self-centred solecism in German national
life. The Coryphaeus was ably supported by a crowd hitherto mute.
The opponents of industrial and the opponents of religious liberalism,
men of rank, men of letters, and high ecclesiastics joined in the
chorus, and another “black day” (July 30, 1878) was added to the
Jewish calendar. In Adolph Stöcker, a Christian Socialist and court
preacher, and a staunch Conservative in the Prussian Diet, the new
crusade found its Peter the Hermit. He was the first man of position
to preach from the pulpit and to declare in the press that Hebrew
influence in the State was disastrous to the Christian section of the
community, that Semitic preponderance was fatal to the Teutonic
race. As though the printing presses of Germany were only waiting
for the signal, a whole library of anti-Semitic literature was rapidly
produced, and as rapidly consumed. Some of the most popular
journals opened their columns to the campaign, Jewish journalists
opposed violence with violence, and the feud daily assumed larger
dimensions, until by the end of 1879 it had spread and raged over
the whole of the empire.
“It is not right that the minority should rule over the majority,”
cried some. Others accused the Jews, loosely and without adducing
any proofs, of forming a freemasonry and of always placing the
interests of their brethren above those of the country. That there was
some kind of systematic co-operation among the Jews seems
probable. It is also probable that there was a certain degree of truth
in the charge of “clandestine manipulation of the press” for the
purpose of shielding even Jews unworthy of protection. But for this
the Germans had only themselves to thank. By attacking the Jews
as a tribe they stimulated the tribal feeling among them. The social
isolation to which they condemned the Jew intensified his gift of
reciprocity. To the German Christians the Jew, however patriotic and
unexceptionable he may be as a citizen, as a man is a Jew—an
alien, an infidel, an upstart, a parasite. His genius is said to be purely
utilitarian, his religion externally an observance of empty forms,
essentially a worship of the golden calf, and worldly success his
highest moral ideal. German professors analysed the Jewish mind
and found it Semitic, German theologians sought for the Jewish soul
and could find none. Both classes, agreeing in nothing else,
concurred in denouncing the Jew as a sinister creature, strangely
wanting in spiritual qualities—a being whose whole existence, devoid

PDF Communications in Interference Limited Networks 1St Edition Wolfgang Utschick Eds Ebook Full Chapter

Uploaded by

Copyright:

Available Formats

You might also like

PDF Communications in Interference Limited Networks 1St Edition Wolfgang Utschick Eds Ebook Full Chapter

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

PDF Communications in Interference Limited Networks 1St Edition Wolfgang Utschick Eds Ebook Full Chapter

Uploaded by

Copyright:

Available Formats

Communications in Interference

Limited Networks 1st Edition Wolfgang

Networks and Communications NetCom2013 Proceedings of

Full Duplex Communications and Networks 1st Edition

Security Aware Device to Device Communications

Intelligent Vehicular Networks and Communications.

Machine Learning and Cognitive Computing for Mobile

Interference Analysis: Modelling Radio Systems for

Big Data Computing and Communications Second

Information and Communications Security 18th

Wolfgang Utschick Editor

ISSN 1860-4862 ISSN 1860-4870 (electronic)

Library of Congress Control Number: 2015950002

Springer Cham Heidelberg New York Dordrecht London

Printed on acid-free paper

Springer International Publishing AG Switzerland is part of Springer Science+Business Media

The classical approaches to cope with interference in wireless communication

computation models, interference alignment, interference limited networks, inter-

Munich, Germany Wolfgang Utschick

1 Coding Techniques for Transmitting Packets Through Complex

8 Distributed Channel Selection for Underlay Device-to-Device

19 Physical Layer Cooperation in Two-Way Relaying Systems . . . . . 451

Martin Bossert, Vladimir Sidorenko and Antonia Wachter-Zeh

Abstract Random Linear Network Coding (RLNC) is a technique to disseminate

Random linear network coding (RLNC) is a technique for transmitting packets

M. Bossert (B) · A. Wachter-Zeh

© Springer International Publishing Switzerland 2016 1

Fig. 1.1 Random linear network [4]

We want to transmit information through a network from a source to a destination.

1F denotes a finite field with q elements, where q is a prime power.

where M is a (n × m) code matrix of a matrix code C M and In is the n × n identity

where M is the transmitted codeword. The matrix L is a known (n × n) matrix such

1.2 Gabidulin Codes

rk(A + B) ≤ rk(A) + rk(B)

and we define the minimum rank distance d of a code C as

An Cn,M,d code over Fq m denotes a code (not necessarily linear) of cardinality M

M ≤ q min{n(m−d+1),m(n−d+1)} = q m(n−d+1) . (1.1)

For linear codes of length n ≤ mGabidulin and dimension k, this implies d ≤

where g j ∈ Fq m , j = 1, . . . , n, are linearly independent over Fq . The notation

A parity-check matrix has the form

where h j ∈ Fq m , j = 1, . . . , n, are linearly independent over Fq , and

where f i ∈ Fq m . The maximum i for which f i = 0 is called the q-degree of f (x)

1.2.1 Decoding Using the Linearized Extended Euclidean

This is equivalent to the following homogeneous linear system of equations:

If we use Gaussian elimination, the complexity of this step is bounded by O(n 3 )

1.2.2 Linearized Shift-Register Synthesis

A shift-register (l, σ ) generates the element sn , if dn (l, σ ) = 0. The shift-register

Output: l, σ (x), and σ (x), n

The problem of linearized shift-register synthesis can be tackled by solving a linear

1.3 Interleaved Gabidulin Codes

1.3.1 Decoding Interleaved Gabidulin Codes

Consider interleaving of identical Gabidulin codes G(q m ; n, k) and index runs

Definition 1.3.1 [32] The -interleaved Gabidulin code is

e = aB, where a ∈ Fqt m , B ∈ Fqt×n ,

and rk(a) = rk(B) = t. We can write B as

where rk(B(l) ) ≤ t. Then we have

e(l) = aB(l) (1.9)

and the syndrome of the l-th code is

where f i ∈ Fq m . The maximum i for which f i = 0 is called the q-degree of f (x)