360 likes | 374 Views
Explore a genetic algorithm solution for network infrastructure planning in surveillance setups to optimize coverage and costs while minimizing additional nodes. Address constraints like capacity, frequency, and geographical limitations.
E N D
Multi-Objective Location - Allocation Planning of Heterogeneous Networks Infrastructure applied to surveillance problem Ons Abdelkhalek, Saoussen Krichen and Adel Guitouni Institut Superieur de Gestion de Tunis Faculty of Law, Economics and Management, Gustavson School of Business, University of Tunis, University of Jendouba University of Victoria, LARODEC lab LARODEC lab Defense Research and Development 0
Outline • Problem Statement • Literature Review • A Multi-Objectives Location - Allocation Planning of Heterogeneous Networks Infrastructure • A Genetic Algorithm for a Multi-Objective Nodes Placement Problem in Heterogeneous Network Infrastructure for Surveillance Applications • Problem Formulation • Solution Approach: Genetic algorithm • Simulator Environment : INFORMLab • Conclusion & Future Work
Motivations • Planning growth and extension of existing networks to avoid the network partitioning and “dead area” • Optimizing nodes placement for both one-hop and multi-hop mode • Creating a new robust architecture integrating and taking advantage of various networking techniques • Multi-platforms communication device • Addressingunexpectedevents by designing contingencies strategies to maximize the reliability of the network
Problem statement • Given: • A set of connected nodes that constitute an initial static networking infrastructure • Anticipated demand distribution • Coverage gaps (dead spots) • A set of candidate sites and test points • Set of communication devices characterized by : cost, power, capacity, range and connection bandwidth • Geographical constraints (mountains, buildings, lake, distances...) • … • Where to position which communication device in order • To optimize coverage, costs and bandwidth • To minimize the number of additional nodes • Under constraints such as capacity, frequency, connection and other physical, environmental and technical constraints (e.g., geographical constraints)
Problem illustration • Given • a distribution of forecasted demand • an initial network infrastructure • a temporal localized demand surge • How to optimize additional networking infrastructure? • How to extend static network with Dynamic or MANETs to address opportunistic demand surge? cj cm Capacity Lead time … c1 R: relay G:Gateway C:Control Cj: Capacity constraints
Literaturereview: Integration architecture in wireless networks • One-hop to one-hop integrating architecture in wireless networks • Antennas Placement problem and Transmitters Placement problem • Heterogeneous transmitters Placement problem • Global planning problem • One-hop to multi-hop integrating architecture in wireless networks • Multi-hop Cellular Network (MCN) • Adaptive Multi-hop Cellular Architecture (AMC) • Ad hoc GSM (A-GSM) architecture • Integrated Cellular and Ad Hoc Relaying system (iCAR) • Hybrid Wireless Network (HWN) Architecture • Muti-hop to multi-hop • Coverage problems in wireless ad-hoc sensor networks
How to plan a new infrastructure? • Different type of wireless structures (e.g., independent (or ad hoc) and infrastructure networks : hierarchical, two tiers..) [A.Bahri and S.Chamberland, 2005] • Heterogeneous resources (e.g., relays, antennas, gateways, controllers. APs, …) • Examples of particular networks include sensor network, cellular network, ad hoc network, surveillance networks… • Frequencies, channels and capacities might be considered in the network planning • Multiple constraints: Physical limitation (e.g., capacity boundary, energy..) , geographic distribution, interferences, capacity constraints…
How to plan a new infrastructure? • Evaluation of networks design is a multicriteria decision problem: • Many conflicting objectives: maximizing coverage, capacity, while minimizing costs and interference for example • Minimizing the number of connecting nodes • Forecasting the demand is another challenge • We suppose though that the demand is known as a starting point over the time horizon.
Initial Thoughts to Address the Planning Problem • Multi-objective optimization problem • min costs, max the quality of services, min the energy consumption, max the coverage, min interferences… • How to estimate the demand? • The demand depends on the time • Define the function of evaluation f(s) (dynamic nodes) • Determine the user models • Maximizing the quality of signalling using metrics (Quality of Coverage, the request blocking, dropping rate..) • Unknown demand based on uncertainty • We can suppose an initial predefined infrastructure • Which resources will remain and which should be changed or removed? • Costs of additional infrastructures if needed
Develop contingencies strategies to include MANETs to extend coverage and services • Extending an existing infrastructure: • Problem under uncertainty Management of the contingency plan • Positioning of relays, antennas, nodes… • Dynamic and multifunction extra-nodes • Capacity, interference, connections, protocol constraints • Extension depends on demand surge, coverage extension needs, time... • Forecasted versus non forecasted • Dynamic decision problem Adaptation strategies
Validate the proposed approach on an empirical case • Consider use cases • Surveillance and mobile platforms • Cellular phone networks • Military applications (Piracy) • Vehicular wireless ad hoc networks • Equip cars with wireless transceivers
Location-Allocation Planning of Heterogeneous Networks Infrastructure • One-hop to Multi-Hop and Multi-Hop to Multi-Hop networks connections • The majority of research efforts focus on possible communication scenarios, technical architecture and routing protocols in an heterogeneous environment • None of the previous works considered multi-objective mathematical model to optimize the infrastructure of an existing network using heterogeneous nodes • The problem is how, what and where to place nodes (heterogeneous nodes)
Problem illustration Range Cost Cost Type • Find the optimal number, position, communication types and connections in a special area of coverage. Power Agents * Communication device 1 * Capacity have * * 1 1 1 Bandwidth Existinginfrastructure 1 cover 1 Candidate sites * * Test points Spatial coordinates Data demand Signal threshold Spatial coordinates
Wide-Area Surveillance Problem A combined operation of many platforms, sensors and communication network systems Optimize the infrastructure in order to allow platforms to communicate between each other Source: APL Technical Digest July-Sept. 2000, Vol. 21, No.3 16
Problem Formulation (1/4) • Maximizing the coverage of the integrated networks Where And • Signal strength between test points and receivers (Ting et al 2009)
Problem Formulation (2/4) • Minimizing costs • Maximizing bandwidth • Constraints: • Agent connection constraints:
Problem Formulation (3/4) • Link connection constraints • Assignment constraints • Each node can have more than one communication device • Each node is assigned to one candidate site • Each candidate site is assigned to at most one node • Each receiver can be assigned to at most one node • Each node should be assigned to at least one other node • At least one node should be connected to the existing networks Z
Problem Formulation (4/4) • Nodes capacity constraints Where • Agents connection capacity constraints • Binary constraints
Investigation of Solution Approaches • Multiple issues: • flexibility in the number of placed nodes, • heterogeneity of nodes, • optimization of multi-objective functions, and • satisfactions of multiple network constraints. • Solution approach: • Meta-heuristics • Evolutionary algorithm: • Each node into the encoded chromosome presents a substring that consists of the position of candidate site where it is located, the communication devices that he is using and the number of other nodes that he is assigned to.
Multi-Objective Genetic Algorithm • Initial population: Randomly generating the number of substring in a chromosome where positions are picked from the set of candidate site, type CD from the input matrix and number of other nodes that are connected to are generated randomly • Fitness evaluation We adopted the NSGA II method for MO problem. Rank and crowding distance depend on a comparison of the objectives. • Selection: We used roulette wheel selection where the probability that a chromosome will be selected is proportional to its fitness. • Crossover: We adopted the one-point crossover operation, where the chromosome is divided into two parts at a random point between substring. Then the two parts are exchanged with each other • Mutation: We adopted the bit-flip mutation where Pm = (1/substring_length),thus each bip has a probability of Pm to be flipped
Representation of chromosome and substring 0 1 N-1 chromosome substring {list of common modes}, if both agents are charring the same network)and the distance between node i and the existing node is < max(range of the two agents) so we connect the two agents . Candidate site index Communication device is assigned to node i or not
Implementation Approach in IL GML points are extracted from Canada's topographical maps of south-east Vancouver Island, the Gulf Islands and part of the Lower Mainland. Inform Lab simulator Iterations
Muti-agent system Analyzing and modeling the problem Mathematical formulation Resolution approach Iterations Meta-heuristique A set of non dominated solutions Implementation run IL simulator
Illustrative Example • The MOGA parameters: • The problem parameters: • Example’s Results:
Random generated instances • Experiments configuration: • Experimental Results:
Empiricalresults • The CPU time is proportional to the problem size and, in average, is about to 2 seconds. As the number of test points is greater in the region of interest, the execution of our optimizer remains longer. • The cost is proportional to the size of the land. We can notice that the more we have test points to cover, the more expensive is the cost of our placement due to communication devices' cost. • In all the problem instances, almost all test points were covered and their demands were satisfied by the new placed nodes. It shows that our MOGA almost converges to the optimal solution. • The number of potentially efficient solution is not really high compared to the generations instances. It can be justified by the small number of candidate sites considered in this problem instances or the communication devices. Other problems should be considered for additional empirical validation.
Conclusion & Future works • INFORMLab vignettes will use our solution in order to optimize nodes placement • Compare different node placement solutions in real-time simulation environment • Future works: • Better integration within INFORMLab • Investigate a combination between meta-heuristics and exact methods (Integration of CPlex) • Test our approach in other environment like cellular wireless networks architecture • Use dynamic node placement with stochastic demand distribution
References • Abderraouf Bahri, Steven Chamberland (2005). On the wireless local area network design problem with performance guarantees. Computer Networks, 48, 856-866 • Ahmed H. Zahran, Ben Liang, Aladdin Saleh (2008). Mobility Modeling and Performance Evaluation of Heterogeneous Wireless Networks. IEEE transactions on mobile computing, 7(8), 1041–1056 • Bharat Bhargava, Xiaoxin Wu, Yi Lu, Weichao Wang (2004). A Cellular Aided Mobile Ad Hoc Network (CAMA). Mobile Networks and Applications, 9, 393-408 • C. Y. Lee and G. H. Kang (2000). Cell planning with capacity expansion in mobile communications: A tabu search approach. IEEE Trans. Veh. Technol., 49(5), 16781691 • Chuan-Kang Ting, Chung-Nan Lee, Hui-Chun Chang and Jain-Shing Wu (2009). Wireless Heterogeneous Transmitter Placement Using Multiobjective Variable-Length Genetic Algorithm. IEEE Transactions on Systems, MAN, and Cybernetics Part B : Cybernetics, 39(4), 945–958 • Dave Cavalcanti, Dharma Agrawal, Carlos Cordeiro, Bin xie and Anup Kumar (2005). Issues in integrating cellular networks, WLANs, and MANETs: A futuristic heterogeneous wireless network. Toward Seamless Internetworking of Wireless LAN and Cellular Networks, IEEE Wireless Communications, 30–41 • Dusit Niyato, Ekram Hossain (2009). Dynamics of Network Selection in Heterogeneous Wireless Networks: An Evolutionary Game Approach. IEEE Transactions On Vehicular Technology, 58(4), 2008–2017 • Fei Yu and Vikram Krishnamurthy (2007). Optimal joint session admission control in integrated WLAN and CDMA cellular networks with vertical handoff. IEEE Transaction on Mobile Computing, 6(1), 126-139 • George Neonakis Agg´elou, Rahim Tafazolli (2001). On the Relaying Capability of Next- Generation GSM Cellular Networks. IEEE Personal Communications, 40–47 • George T. Karetsos, Sofoklis A. Kyriazakos, Evangelos groustiotis, Felicita Di Giandomenico, Ivan Mura (2005). A hierarchical radio resource management Framework for integrating WLANs in Cellular networking environments. Internetworking wireless LAN and Future cellular networks 12(6), 11–17 • Hongyi Wu, Chunming Qiao, Swades De, Ozan Tonguz (2001). An integrated cellular and ad hoc relaying system: iCAR. IEEE Journal on Selected Areas in Communications, 19(10), 2105-2115 • Hongyi Wu, Chunming Qiao, Swades De, Evsen Yanmaz, Ozan Tonguz (2005). Quality of Coverage (QoC) in Integrated Heterogeneous Wireless Systems. Springer-Verlag Berlin Heidelberg, 689-700 • Hongyi Wu, Swades De, Chunming Qiao, Evsen Yanmaz, Ozan K. Tonguz (2005). Hand-Off Performance of the Integrated Cellular and Ad Hoc Relaying (iCAR) System. Wireless Networks, 11, 775-785 • S. Toumpis, D. Toumpakaris (2006). Wireless ad hoc networks and related topologies: applications and research challenges. Elektrotechnik& Informationstechnik, 123(6), 232-241 • Wei Song, Hai Jiang, Weihua Zhuang (2007). Performance analysis of the WLAN first scheme in cellular • S.Meguerdichian, F. Koushanfar, M.Potkonjak, M.B.Srivastava (2001). Coverage problems in wireless ad-hoc sensor networks. INFOCOM 2001:Twentieth Annual Joint Conference of the IEEE Computer and Communications Societies. Proceedings. IEEE. 3. 1380-1387