Résumé:
This research endeavors to enhance the efficiency of Wireless Sensor Networks (WSNs)
deployment specifically tailored for Smart Car Parks (SCP) surveillance applications. Traditional
deployment methods often follow a sequential approach, leading to suboptimal solutions
in terms of node placement, network coverage, and connectivity. To address these limitations,
our proposed approach advocates the simultaneous placement of Sensor Nodes (SNs) and Relay
Nodes (RNs) to minimize the overall number of required nodes while ensuring sufficient
coverage and connectivity throughout the car park area.
The core of our methodology lies in the development of a novel optimization framework
that integrates both deterministic and meta-heuristic techniques. Firstly, we introduce a Multi-
Objective Linear Programming model (MOLP) capable of efficiently solving smaller instances
of the deployment problem, optimizing objectives such as the number of SNs, number of RNs,
and network diameter simultaneously. To tackle larger instances of the deployment problem,
we complement MOLP with the Greedy Chaos Whale Optimization meta-heuristic (GCWOA),
which is designed to handle complex optimization scenarios effectively.
GCWOA combines several optimization strategies to achieve superior deployment solutions.
Initially, a Greedy algorithm is employed for the initial placement of nodes, facilitating a quick
but reasonably good solution. Subsequently, a Chaos Local Search (CLS) technique is integrated,
leveraging chaos maps to explore the solution space effectively and refine the initial placement.
Furthermore, the meta-heuristic divides the solution population into two sub-populations, each
subjected to different optimization techniques. One sub-population undergoes CLS for further
refinement, while the other utilizes the Whale Optimization Algorithm (WOA) to exploit the
search space more intensively, thereby enhancing the solution quality.
An extensive comparative performance evaluation against established meta-heuristics such
as the basic WOA, Genetic Algorithm (GA), and Particle Swarm Optimization (PSO) demonstrate
the superiority of GCWOA. On average, GCWOA achieves a significant fitness improvement
of 23.45%, 28.75%, and 26.32% compared to basic implementations of WOA, GA, and
PSO, respectively. Moreover, GCWOA exhibits faster convergence rates and reduced runtimes,
highlighting its effectiveness in tackling SCP surveillance deployment challenges efficiently and
reliably.
