Deploying m-connected k-covering (MK) wireless sensor networks (WSNs) is crucial for reliable packet delivery and target coverage. It has always been a challenge to minimize the size of such networks to lower the deployment costs. However, this problem, also known as the minimum m-connected k-coverage problem, is NP-complete. This paper proposes a novel approach, called DCPMK, based on difference-of-convex programming (DCP) to solve the minimum m-connected k-coverage problem. Given an initial MK WSN, the idea is to iteratively shrink the inter-node distances by solving a proposed DCP problem and then exclude the redundant nodes. The proposed method is applicable to both 2D and 3D environments, and it guarantees both m-connectivity and k-coverage properties. Simulations reveal the considerable superiority of DCPMK over several leading metaheuristic-based approaches in terms of network size, computational time overhead, and network lifetime. Specifically, DCPMK outperforms the benchmark methods for large fields of interest and a greater number of target points, hence demonstrating considerable scalability.
