Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

University of Babylon

College of Information Technology


Dept. of Information Networks
Data Communication and Networking II.
4th Stage- Lecture 8.

Lecture 8

Named Data Networking (NDN)

1 Introduction
Information-centric networks (ICNs) introduce a radical change in Internet communications. ICNs
emphasize information access regardless of location through a new data-based approach, allowing
networks to actively deliver content. The more interesting project that draw attention to the
notion of information centric networking are the Content Centric Networking (CCN) project and
Named Data Networking (NDN).

1.1 Content naming


In current IP based networking, content retrieval over the Internet is intrinsically host-centric,
which implies knowing the host IP address to send content requests to. A priori knowledge of an
end system IP address is required to establish one or more communication paths between the user
and content host and to allow content to be directly requested. This host-centric approach binds
content to identification and location. The ICN approach differs significantly from traditional
host-centric network architectures. By treating content rather than its storage node as a first-class
citizen in the network architecture, ICNs allow content request and retrieval to occur in a name-
based manner. This can be accomplished by using specific content naming schemes. An ideal
naming scheme should provide names with the following set of features:

• Uniqueness: to assure unique content identification with the smallest possible false positives
and negatives.
• Persistence: to guarantee the validity of the content name in synchronization with the content
validity itself.
• Scalability: to allow different namespace scales, serving tiny and huge namespaces the same
with no limitations regarding content nature, storage location or any other characteristic.

1.2 Content or name-based routing


Unlike host-centric networks, ICNs should be able to deliver content requested by name, without
any information regarding its storage location. To this end, ICN nodes need information about
existing content on the network to efficiently route content requests to valid copies of that content.
This content or name-based routing should have the following characteristics:

• Content oriented: packets should be addressed to content names, without information about
source or destination.

• Robustness: routing should be fault tolerant and quickly recover from discontinuities, avoid-
ing sending data to faulty nodes.
• Efficiency: control information should have low impact on network traffic.

1
• Scalability: routing should be flexible so as to be deployed in a variety of scenarios, serving
dense network topologies and large namespaces and tiny local networks as well.

2 NDN Architecture
NDN is a receiver-driven, network-layer communication protocol. This section describes a brief
overview of the NDN paradigm.

2.1 Packets and Names


The entire communication in NDN is accomplished by utilizing two various kinds of packets:
Interest and Data (see Figure 1). These packets hold an NDN name that uniquely identifies a
chunk of data that can be transferred in one Data packet

Figure 1: NDN packet types.

To obtain a Data, a client inserts the requested data name into an Interest packet and forwards it to
the network. This name is utilized by routers to forward the Interest towards the data provider(s),
and the Data packet whose name matches the name in the Interest is fetched back to the client.
Moreover, every Interest packet holds a selector field which gives extra particular specifications of
the required Data, and a nonce field which is a random number produced by the client. A Data
packet holds the requested data, data characterization, and cryptographic signature that connects
the data to the name. Data providers signed the Data packets when they generated them. The
consumers utilizes the signature to investigate the probity of Data packets. Consequently, Data
packet confidence is separated from how it is gained or what its source. The security of NDN’s
built-in data-centric is important in different sides. As an instance, signatures on network routing
and control messages give a strong basis for securing routing protocols against attacks like tam-
pering and spoofing.
NDN names are ambiguous to the network (i.e. routers do not realize what they mean) and
this gives the ability to the applications to determine their own naming approaches autonomous
from the network.Like URL names, NDN names are hierarchically structured. As an instance,
/iq/edu/university/Babylon/, where ’/’ is the delimiter between name components. Two benefits
for this hierarchical naming structure. First, applications can put the context and relationship
among data elements into the names. As an instance, partition two of version three of paper1 in
Babylon university can be named as: /edu/Babylon/Computerengineering/paper1.pdf/3/2. Sec-
ond, it permits for name aggregation which is primary in scaling the routing system. As an
example, let University of Babylon be an independent system in NDN, it could disseminate the
name prefix: /edu/ Babylon/ through routing protocols in a manner comparable to disseminating
IP prefixes in present Internet.

2
2.2 Forwarding Process
Every NDN router preserves three essential data structures to perform packet (Interest and Data)
forwarding processes. These data structures are: Forwarding Information Base (FIB), Pending
Interest Table (PIT), and Content Store (CS) (see Figure 2). FIB table in NDN platform is
comparable to the FIB in IP platform that it holds name prefixes rather than IP address prefixes,
and it may show many interfaces for a specific name prefix. As its name refer, the PIT saves the
entire Interest packets that have been forwarded but not yet satisfied. Every PIT entry registers
both the Interest packet name and the incoming interface(s). The CS provides provisional storage of
the fetched Data packets. Also, every NDN router is supported with a forwarding strategy, which
specifies whether, when and where to forward an Interest packet depending on the information
saved in these data structures.

Figure 2: Interest and data processing in NDN.

Whenever NDN router receives an Interest packet, first it will examine whether there is a matching
Data in its Content Store and, if so, it will respond immediately by sending the data back through
the interface from which the Interest is received. Otherwise the Interest packet name will be
examined against PIT entries. If a match exist, only the incoming interface will be recorded in the
matched PIT entry. If there is no match with PIT entries, a new PIT entry will be added and both
the incoming Interest packet name and the incoming interface will be recorded in this new entry.
Then, this Interest packet will be forwarded depending on the applied forwarding strategy and the
information registered in the FIB table. If at any time a Data packet received by NDN router,
its name is examined against PIT entries. If a match exist, this Data packet will be forwarded
through the interface(s) that had been recorded in this PIT entry and, then, this PIT entry will be
removed. Also, it is possible to save a copy of this data in the CS (depending on its caching policy)
to satisfy the future requests for this Data packet. Consequently, Data packets permanently take
the inverse paths of Interest packets. If there is no match, this Data packet will be ignored. Every
PIT entry has an associated lifetime (comparable to the TTL field in the IP architecture) and it
is utilized to prevent a PIT entry from being exist for unlimited time. The value of this lifetime is
specified by the consumer and a PIT entry is deleted if the Interest not satisfied before its lifetime
expires. Since a Data packet holds a name and a signature, it is significative autonomous of its
source and destination. Consequently, Data packets cached in the CS can be utilized to satisfy
future Interests. CS is similar to memory buffer in IP routers, but IP routers cannot use a packet
again after it is forwarded. In-network caching is beneficial in multiple situations. As an instance,
a Data packet may get dropped by the network because of congestion after being cached by router

3
A. When the consumer retransmits the Interest, it can retrieve the Data immediately from router
A without being additionally forwarded to the genuine provider.
Figure 3 shows an example of retrieving content in NDN paradigm (the numbers on arrows depict
the sequence of events). In this example, firstly Rose sends the Interest packet. When the Interest
packet sent from Rose reaches at the gateway router, the router first examines the content store,
which caches the already requested Data packets, and checks if the requested content is already
there. In this example, there is no such content in the CS, so the router, next, examines the PIT to
check if there is already a request for this content. Also, there is no such request in our example.
Consequently, the Interest packet is forwarded depending on the FIB lookup results, and a new
entry, which contains the incoming interface and the name of the requested content, is inserted in
the PIT. When the Interest packet arrives at the content provider, the Data packet is sent back.
Whenever the Data packet reaches at the gateway router, the router examines whether a pending
PIT entry is waiting for this Data packet. If so, the Data packet is sent back to the incoming
interface(s) and the PIT entry is eliminated. Next, the Data packet is cached in the CS. Later,
when John sends the Interest packet, there is a CS hit at the gateway router. Consequently, the
Data packet is delivered from the gateway router to John directly. It can be noticed that the
requested content is sent from the content provider only once in this example.

Interest Packet
Data Packet

Rose

2 3
5 4

Content Provider

John

Figure 3: NDN content delivery example.

You might also like