Node Placement and Path Planning for Improved Area Coverage in Mixed Wireless Sensor Networks

TL;DR

The paper tackles area coverage in mixed WSNs by jointly optimizing static node placement and mobile-node path planning using three MILP formulations: MILP-Static places static sensors with emphasis on boundary coverage, MILP-Cov drives mobile paths to maximize coverage within a movement limit, and MILP-Mov minimizes movements to achieve a target coverage. The approach leverages a grid-based model with variables for node locations $x^{s}_{i,j}$, $x^{l,k}_{i,j}$ and coverage indicators $c^{s}_{i,j}$, $c^{l,k}_{i,j}$, $c_{i,j}$, under mobility bounds $(\rho_x,\rho_y)$ and sensing radius $r_s$, and enforces a coverage ratio $cr$. Extensive simulations on $M\times N$ grids demonstrate that MILP-Static significantly improves boundary coverage; MILP-Cov and MILP-Mov achieve higher area coverage with fewer mobile movements than benchmark methods, including Vecchio2015. The results show linear growth in problem size and that MILP-Cov typically has lower runtime than MILP-Mov, with static placement enabling more effective path planning and potential lifetime extensions for the network.

Abstract

For the large-scale monitoring of a physical phenomena using a wireless sensor network (WSN), a large number of static and/or mobile sensor nodes are required, resulting in higher deployment cost. In this work, we develop an efficient algorithm that can employ a small number of static nodes together with a set of mobile nodes for improved area coverage. An efficient deployment of static nodes and guided mobility of the mobile nodes is critical for maximizing the area coverage. To this end, we propose three mixed integer linear programming (MILP) formulations. The first formulation efficiently deploys a set of static nodes and the other two formulations plan the path of a set of mobile nodes so as to maximize the area coverage and minimize the total number of movements required to achieve the desired coverage. We present extensive performance evaluation of the proposed algorithms and its comparison with benchmark approaches. The simulation results demonstrate the superior performance of the proposed algorithms for different network sizes and number of static and mobile nodes.

Figures (5)

• Figure 1: A typical network area. The colored cells around a sensor node indicate the sensing region of that sensor node ($r_s=1$). The cells marked with diagonal lines around each mobile node represent potential locations where the mobile node could move to in the next iteration ($\rho_x=\rho_y=2$).
• Figure 2: Possible challenges due to random deployment of static sensor nodes in a network area. $(a)$, $(b)$ network partition, $(c)$ boundary coverage hole, and $(d)$ overlapping/redundant coverage.
• Figure 3: Area coverage using three $(L=3)$ mobile nodes with different number of static nodes.
• Figure 4: Number of movements required by mobile nodes for full area coverage $({cr = 1})$ using MILP-Mov with different number of static and mobile nodes.
• Figure 5: Area coverage as a function of the number of movements by the mobile nodes (Network size = $10 \times 10$).