PSO Based Topology Optimization for Underwater Acoustic Sensor Network

Underwater Acoustic Sensor Networks (UASNs) are capable of obtaining Marine information effectively by deploying surface gateway node and underwater sensor node to form a three-dimensional wireless sensor network covering the surface to the seabed. Network topology is one of the key problems in the design of underwater acoustic communication network, which plays a decisive role in the capacity, energy consumption and reliability of the communication network. Therefore, it is necessary to generate a better topology structure to improve the performance of the network.

The current research on wireless network topology optimization is more mature than underwater acoustic network, so this paper refers to the topology optimization algorithm of wireless communication network.

In this paper, a Particle Swarm Optimization (PSO) based topology optimization algorithm of UASNs is proposed, which takes connectivity and coverage rate as constraints, takes propagation loss of sound field and operational lifetime of the network as objective functions, and optimizes the topology structure of underwater acoustic network.

Particle Swarm Optimization (PSO) was proposed in 1995 by American scholar Kennedy et al. The algorithm is an intelligent optimization algorithm that simulates the group intelligent behavior such as bird foraging, that is often used in the topology optimization of wireless communication networks. The basic idea of Particle Swarm Optimization algorithm is to search for the optimal solution according to individual and group behavior. This paper tries to apply the algorithm to the topology optimization of underwater acoustic sensor networks. The flow chart is shown as Figure 1.

Constraint conditions are set so that the positions of nodes to be optimized are generated within the specified range at first to ensure communication between the two nodes and a certain coverage rate. Then, the propagation loss of sound field and operational lifetime of the network are taken as objective functions. Weights are set for the two objective functions according to AHP-EWM method, and fitness values are used to represent the objective function values. Namely, <Formula format="inline"><TexMath><?TeX $fitness = \alpha \cdot TL + \beta \cdot D$?></TexMath><File name="a00--inline1" type="gif"/></Formula>. Where the <Formula format="inline"><TexMath><?TeX $TL$?></TexMath><File name="a00--inline2" type="gif"/></Formula>represents the propagation loss of sound field and the <Formula format="inline"><TexMath><?TeX $D$?></TexMath><File name="a00--inline3" type="gif"/></Formula> represents the operational lifetime of the network, <Formula format="inline"><TexMath><?TeX $\alpha,\beta $?></TexMath><File name="a00--inline4" type="gif"/></Formula> represent the weight of <Formula format="inline"><TexMath><?TeX $TL$?></TexMath><File name="a00--inline5" type="gif"/></Formula> and <Formula format="inline"><TexMath><?TeX $D$?></TexMath><File name="a00--inline6" type="gif"/></Formula> respectively. Since the determined objective functions are the smaller the better, the PSO algorithm will be used for minimum optimization.

In order to verify the feasibility of the proposed method, MATLAB is used in shallow sea channel simulation experiments. Topology optimization is carried out on the network with different number of nodes (the number of nodes is 5, 15, 25, 35, 45, respectively). And the output topology structures are shown as Figure 2.