Professional Documents
Culture Documents
Metaheuristics For Resource Deployment Under Uncertainty in Complex Systems 1St Edition Shuxin Ding Online Ebook Texxtbook Full Chapter PDF
Metaheuristics For Resource Deployment Under Uncertainty in Complex Systems 1St Edition Shuxin Ding Online Ebook Texxtbook Full Chapter PDF
https://ebookmeta.com/product/optimal-operation-of-integrated-
multi-energy-systems-under-uncertainty-1st-edition-wu/
https://ebookmeta.com/product/design-of-deep-braced-excavation-
and-earth-retaining-systems-under-complex-built-environment-
theories-and-case-studies-1st-edition-zhang/
https://ebookmeta.com/product/the-science-and-management-of-
uncertainty-dealing-with-doubt-in-natural-resource-
management-1st-edition-bruce-g-marcot/
https://ebookmeta.com/product/large-scale-system-analysis-under-
uncertainty-with-electric-power-applications-1st-edition-
dominguez-garcia/
Methods of Mathematical Modelling and Computation for
Complex Systems Jagdev Singh
https://ebookmeta.com/product/methods-of-mathematical-modelling-
and-computation-for-complex-systems-jagdev-singh/
https://ebookmeta.com/product/ai-metaheuristics-for-information-
security-in-digital-media-1st-edition-apoorva-s-shastri/
https://ebookmeta.com/product/metaheuristics-for-finding-
multiple-solutions-1st-edition-mike-preuss/
https://ebookmeta.com/product/multiscale-dynamics-simulations-
nano-and-nano-bio-systems-in-complex-environments-1st-edition-
salahub/
Metaheuristics for
Resource Deployment
under Uncertainty
in Complex Systems
Metaheuristics for
Resource Deployment
under Uncertainty
in Complex Systems
Shuxin Ding
Chen Chen
Qi Zhang
Bin Xin
Panos M. Pardalos
First edition published 2022
by CRC Press
6000 Broken Sound Parkway NW, Suite 300, Boca Raton, FL 33487-2742
© 2022 Shuxin Ding, Chen Chen, Qi Zhang, Bin Xin, Panos M. Pardalos
Reasonable efforts have been made to publish reliable data and information, but the author and
publisher cannot assume responsibility for the validity of all materials or the consequences of their use.
The authors and publishers have attempted to trace the copyright holders of all material reproduced
in this publication and apologize to copyright holders if permission to publish in this form has not
been obtained. If any copyright material has not been acknowledged please write and let us know so
we may rectify in any future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced,
transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or
hereafter invented, including photocopying, microfilming, and recording, or in any information
storage or retrieval system, without written permission from the publishers.
For permission to photocopy or use material electronically from this work, access www.copyright.com
or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923,
978-750-8400. For works that are not available on CCC please contact mpkbookspermissions@tandf.
co.uk
Trademark notice: Product or corporate names may be trademarks or registered trademarks and are
used only for identification and explanation without intent to infringe.
DOI: 10.1201/9781003202653
Typeset in LatinModern
by KnowledgeWorks Global Ltd.
Contents
Preface xi
Acknowledgments xv
Chapter 1 Introduction 1
v
vi Contents
2.1 INTRODUCTION 29
2.2 PROBLEM FORMULATION 31
2.2.1 Detection Models 31
2.2.1.1 Binary Detection Model 31
2.2.1.2 Probabilistic Detection Model 32
2.2.2 Network Model 32
2.2.3 Problem Statement 33
2.2.4 NP-Hardness Proof 34
2.3 SOLUTION ALGORITHMS 35
2.3.1 D-VFCPSO 35
2.3.2 Other PSO-Based Algorithm for Area Coverage
Problem 38
2.3.3 Complexity Analysis 39
2.4 EXPERIMENTS AND DISCUSSION 39
2.4.1 Test Instances 40
Contents vii
3.1 INTRODUCTION 46
3.2 PROBLEM FORMULATION 47
3.2.1 Mathematical Model 49
3.2.2 Scenario-Based Model Reformulation 49
3.3 SOLUTION ALGORITHMS 50
3.3.1 NSGA-II 50
3.3.2 MOPSO 52
3.3.2.1 Personal Best Selection 52
3.3.2.2 Non-Dominated Solutions Maintaining
and Global Best Selection 53
3.3.2.3 Diversity Maintaining 53
3.3.3 Complexity Analysis 55
3.4 EXPERIMENTS AND DISCUSSION 56
3.4.1 Test Instances 56
3.4.2 Performance Metrics 57
3.4.3 Parameter Turning 59
3.4.4 Analysis of Results 60
3.5 CONCLUSION 74
4.1 INTRODUCTION 76
4.2 PROBLEM FORMULATION 78
4.2.1 The Deterministic and Uncertain Two-Level
Cooperative Set Covering Problem 78
viii Contents
Bibliography 171
Index 189
Preface
xi
xii Preface
Shuxin Ding
Chen Chen
Qi Zhang
Bin Xin
Panos M. Pardalos
Acknowledgments
xv
Author Bios
Shuxin Ding received the B.E. degree in automation and the Ph.D. de-
gree in control science and engineering from the Beijing Institute of Tech-
nology, Beijing, China, in 2012 and 2019, respectively. He is currently
an assistant researcher with the Signal and Communication Research
Institute, China Academy of Railway Sciences Corporation Limited. His
current research interests include railway scheduling, evolutionary com-
putation, multi-objective optimization, and optimization under uncer-
tainty.
Chen Chen received the B.S. degree in automation and the Ph.D. de-
gree in control science and engineering from the Beijing Institute of
Technology, Beijing, China, in 2004 and 2009, respectively. She is cur-
rently a professor with the School of Automation, Beijing Institute of
Technology. Her current research interests include complicated systems,
multi-objective optimization, and distributed simulation.
Bin Xin received the B.S. degree in Information Engineering and Ph.D.
degree in Control Science and engineering, both from the Beijing Insti-
tute of Technology, Beijing, China, in 2004 and 2012, respectively. He
xvii
xviii Author Bios
Introduction
Resource deployment problem analyzes how to set the locations for de-
ploying resources with the best performance and lowest cost. The re-
sources can be static nodes and moving nodes. These resources can
provide services for a specific area or some customers. This chapter in-
troduces the resource deployment problem. It provides some real-world
applications and fundamental issues of node deployment problem, and
research progress of node deployment modeling and methods. This chap-
ter provides a basic foundation for the whole work and gives the main
issue and challenges that lead to the following chapters.
DOI: 10.1201/9781003202653-1 1
2 Metaheuristics for Resource Deployment
:DWHU
1.1.3 Healthcare
Healthcare is defined as the prevention and treatment for illness or injury
through professional medical services. The facility location problems re-
lated to healthcare covers from locating healthcare facilities to layout
problems in hospitals [16]. Figure 1.3 shows the locations of historical
cardiac arrests and candidate sites for AED deployment. A brief intro-
duction and example applications of healthcare are presented as follows.
Introduction 5
+LVWRULFDOFDUGLDFDUUHVWV
&DQGLGDWHVLWHV
Figure 1.3
Locations of historical cardiac arrests and candidate sites for
AED deployment.
&KHFNRXW
DELNH
&KHFNLQ
DELNH
%LNHVKDULQJ VWDWLRQV
+LJKVSHHGUDLOZD\VWDWLRQV
5DLOZD\OLQHV
load among them. The results of load balancing are shown in the fol-
lowing aspects: shortening the average response time of tasks effectively,
maximizing the use of simulation resources of the whole system, and
reducing the unnecessary waste of resources [26]. During the process of
simulation, additional federates or components will be added to complete
specific simulation tasks. As shown in Figure 1.6, when the simulation
process is advanced to step k, a certain number of federates or compo-
nents must be added. The key to solving this dynamic load balancing
problem is transferring dynamic states to static states, converting the
problems into static load balancing problems.
1.2.1 Task
Optimizing the resource deployment problem is a critical issue in com-
plex systems. Heterogeneous resources, e.g., sensors and actuators, can
work cooperatively through the network, improving the system’s perfor-
mance. It has been applied in the network-centric systems with sensors
and weapon systems, which significantly improve the performance in
air defense systems [29]. In unmanned systems, both UAVs and UGVs
can act as sensor nodes or actuator nodes. Sensor nodes usually per-
form monitoring and surveillance tasks. Actuator nodes perform some
control tasks, e.g., weapon nodes perform attacking towards the enemy
targets. For example, by deploying UAVs and UGVs, surveillance tasks
on a specific area and attacking tasks on the targets within the area are
conducted. The sensor nodes’ environmental information and warning
information can be transmitted to the actuator (weapon) nodes through
the network. Under the guidance of sensor nodes, weapon nodes could
perform effective attacking.
Introduction 11
1.2.2 Node
In real-world applications, there are usually different kinds of nodes
deployed in a complex system. These nodes are deployed to comple-
ment each other’s advantages. However, these nodes are homogeneous
resources, which are belong to the same type, i.e., sensor node or actu-
ator node. Few studies consider the joint deployment of heterogeneous
resources, e.g., sensor and actuator nodes. Therefore, the cooperation
between heterogeneous resources and rational utilization of them needs
further investigation.
1.2.3 Environment
The node deployment problem can also be divided into static deploy-
ment and dynamic deployment. Most of the previous studies analyze
the static deployment problem. Nodes are deployed before any possible
related tasks. It is an off-line node deployment problem without any time
limit. However, when the states of the environment and nodes change
during the service, the deployment’s performance may not meet the fu-
ture requirements. Therefore, the nodes’ location should be adjusted by
redeployment to meet the future requirement.
Besides, most of the studies are deterministic, which means the pa-
rameters of the environment, resources, and targets are determined val-
12 Metaheuristics for Resource Deployment
method can effectively reduce the unnecessary search space and make
the search more efficient. Besides squared grids and sectors, other dis-
cretization methods such as the triangular grid and hexagonal grid are
usually adopted in coverage problems in WSNs [36].
Since the discretization of the space will cause missing some deploy-
ment information inevitably, we should carefully select the size of the
grid. While in the continuous space-based deployment, nodes can be
deployed at any place in the predefined region, which ensures that the
optimal solution must be in the search space. However, the searching
cost of this method is higher than the discrete space-based deployment.
In Refs. [43, 44], the virtual force algorithm is adopted to deal with the
issues in the continuous space-based deployment. The total forces from
other nodes, protected objects, and boundaries are used to adjust the
locations of the nodes. It prevents the nodes from congesting in a local
area and keeps the nodes in a certain density with avoidance of blind
area.
1.3.2 Constraints
The resource deployment problem is a constrained optimization problem.
The constraints limit the range of solutions and ensure that the plan-
ning results are reasonable and effective. In the real-world deployment
problem, the feasible solution is limited by the deployment space, and
accurate geographic information is analyzed to determine the deploy-
able and non-deployable areas. For example, lakes, depressions, forests,
or locations where the slope is too large or the terrain is too low are not
suitable for deploying nodes. It is often difficult to mathematically de-
scribe non-deployable areas (especially irregular non-deployable areas).
Such constraints will be dealt with in a specific manner during the op-
timization process. Except for the constraints in the deployment space,
other features and demands of the nodes should be considered.
The number of the nodes is the first consideration when deploying
nodes. The number of available nodes in different areas is set as con-
straints. The nodes’ cost can also be a constraint since some of the
nodes may be very expensive. In Ref. [51], the coverage demands for
some important directions are set as constraints. Besides, the minimum
distance between nodes is set as a constraint to prevent electromagnetic
interference. Meanwhile, the distance between nodes and the protected
objects should satisfy some requirements. These are typical constraints
in the case of static node deployment.
In dynamic node deployment, the region for deployment changes ac-
cording to the nodes and targets. As a result, the corresponding de-
ployment space and constraints should be adjusted. In Ref. [44], some
deployed nodes are malfunctioning, and the corresponding redeployment
problem needs to be solved based on the updated constraints.
16 Metaheuristics for Resource Deployment
where X ⊆ Rn denotes the set of all feasible solutions. fi (x) denotes the
ith performance metrics for evaluating the deployment. hk (x) denotes
the kth constraint for deployment space requirement. gk (x) denotes the
kth other constraint. x denotes the decision variable, which is a vector
x = [x1 , x2 , ..., xn ]. In the discrete space-based deployment, the candi-
date location for deployment is denoted as p = [p1 , p2 , ..., pm ], where xi
denotes the ith node assigned with the jth location p. In the continuous
space-based deployment, xi denotes the position of the ith node.
The critical issue is how to describe the performance of the nodes
in different locations. Therefore, the objective function can be used to
evaluate the performance of the deployment. There are many different
types of node deployment problems, which involve different optimiza-
tion goals. We mainly consider the objective functions in WSNs and air
defense.
The ability of the deployed resources and the condition of the protected
resources should be analyzed. Meanwhile, the moving direction, position,
and number of targets should also be considered. Similar to the area
coverage and target coverage in WSNs, two types of the node deployment
problem in air defense are usually studied, which are area air defense and
point air defense.
• Area air defense.
Area air defense represents that the entire area needs to be pro-
tected by the deployed nodes (e.g., weapon nodes) [55]. Since the
resources are limited compared with the protected area, it is nec-
essary to select several important sampling points for protected re-
sources to evaluate the deployment performance. In Ref. [55], some
important protected resources are chosen from the entire area and
are used for evaluation. The selection for the protected resources is
similar to the discretization method in candidate locations. Both
uniform and non-uniform sampling points are used for calculation.
To maximize the protection of these resources is belongs to the
coverage problem. The strength of the coverage is more suitable
for performance evaluation in area air defense.
Some studies consider the set covering problem (SCP) and maxi-
mal covering location problem (MCLP) to formulate this problem
and proposed a 0-1 programming problem [20, 56]. SCP minimizes
the deployment cost with satisfying coverage, while MCLP maxi-
mizes the coverage of the targets with limited resources. Ref. [57]
proposes firing range covering and firing angle covering based on
SCP. In Ref. [39], ship sector locations are optimized to provide
a robust air defense formation as the sector allocation problem
(SAP). The problem is formulated as a 0-1 integer linear program-
ming (ILP) problem, and the coverage provided to the ships is
maximized. This is a classic MCLP. Since there are always multiple
covering situations for covering the protected resources in the area
air defense, an aggregate signal function can be used to describe
the cooperative coverage1 [58]. The followings are two methods to
deal with it.
1.4.1 Encoding
Encoding is applied to describe the locations for the deployed nodes.
It is the decision variable and related to the deployment space. There
are differences between discrete space-based deployment and continuous
space-based deployment. In discrete space-based deployment, the deci-
sion variable is usually set as an integer representing whether a resource
has been deployed. Branch-and-bound [39,66] and genetic algorithm [55]
are applied to solve the problem. In continuous space-based deployment,
any position in the deployment space can be the location for deployment.
This encoding mechanism is usually adopted in the problem solving by
particle swarm optimization [32] and virtual force algorithm [43].
1.4.4 Algorithms
Exact algorithms are applied in node deployment, e.g., enumeration
method [72], branch-and-bound [39], etc. Besides, since node deploy-
ment is belongs to NP-hard problem, metaheuristic algorithms, e.g., ge-
netic algorithm [55,59], particle swarm optimization algorithm [50], sim-
ulated annealing algorithm [73], etc. are also applied. These algorithms
are nature-inspired and have a stochastic searching mechanism.
Bodmer went to England for the first time in 1816 and visited all
the principal machine shops, textile mills and iron works. He returned
in 1824 and again in 1833, this time remaining many years. On his
second trip he established a small factory for the manufacture of
textile machinery at Bolton, in which was one of the first, if not the
first, traveling crane.[75] At the beginning of his last and long
residence in England, Bodmer appointed Sharp, Roberts &
Company makers of his improved cotton machinery, which they also
undertook to recommend and introduce. This arrangement was not
successful, and a few years later, in partnership with Mr. H. H. Birley,
Bodmer started a machine shop and foundry in Manchester for
building machinery.
[75] Ibid., p. 581.
Nearly all of the machinery for the Manchester plant was designed
and built by Bodmer himself and it forms the subject of two
remarkable patents, granted, one in 1839 and the other in 1841.[76]
The two patents cover in reality nearly forty distinct inventions in
machinery and tools “for cutting, planing, turning, drilling, and rolling
metal,” and “screwing stocks, taps and dies, and certain other tools.”
“Gradually, nearly the whole of these tools were actually constructed
and set to work. The small lathes, the large lathes, and the planing,
drilling, and slotting machines were systematically arranged in rows,
according to a carefully-prepared plan; the large lathes being
provided, overhead, with small traveling cranes, fitted with pulley-
blocks, for the purpose of enabling the workmen more economically
and conveniently to set the articles to be operated upon in the lathes,
and to remove them after being finished. Small cranes were also
erected in sufficient numbers within easy reach of the planing
machines, &c., besides which several lines of rails traversed the
shop from end to end for the easy conveyance on trucks of the parts
of machinery to be operated upon.”[77] There were, in addition to
these, however, “a large radial boring machine and a wheel-cutting
machine capable of taking in wheels of 15 feet in diameter, and of
splendid workmanship, especially in regard to the dividing wheel,
and a number of useful break or gap-lathes, were also constructed
and used with advantage. It is especially necessary to mention a
number of small, 6-inch, screwing lathes, which, by means of a
treadle acting upon the driving gear overhead, and a double slide-
rest—one of the tools moving into cut as the other was withdrawn,—
screw cutting could uninterruptedly proceed both in the forward and
in the backward motion of the toolslide, and therefore a given
amount of work accomplished in half the time which it would occupy
by the use of the ordinary means. Some of the slide-lathes were also
arranged for taking simultaneously a roughing and finishing cut.”[78]
[76] The first of these is described in the American Machinist of March
13, 1902, p. 369.
[77] “Memoir,” p. 588.
[78] Ibid., p. 597-598.
The latter part of Bodmer’s life was spent in and near Vienna,
working on engines and boilers, beet sugar machinery and
ordnance; and at Zurich, where he died in 1864, in his seventy-ninth
year.
Bodmer does not seem to have originated any new types of
machine tools, with the exception of the vertical boring-mill, which he
clearly describes, terming it a “circular planer.” It was little used in
England, and has been considered an American development.
It is hard now to determine how far Bodmer has influenced tool
design. It was much, anyway. Speaking of the patent just referred to,
John Richards, who has himself done so much for tool design, says,
“Here was the beginning of the practice that endured.” He has
described some of Bodmer’s tools in a series of articles which show
a standard of design greatly in advance of the practice of his time.[79]
Another writer says of Bodmer, “He seems always to have
thoroughly understood the problems he undertook to solve.” “One is
lost in admiration at the versatility of the inventive genius which could
at any one time—and that so early in the history of machine design
—evolve such excellent conceptions of what was needed in so many
branches of the mechanics’ art.”[80]
[79] American Machinist, Vol. XXII, pp. 352, 379, 402, 430, 457, 478,
507, 531, 559, 586, 607, 637.
[80] Ibid., Vol. XXV, p. 369.