Minimizing the Maximum Delay in Wireless Sensor Networks by Intelligent Sink Placement

Research activity in the area of wireless sensor networks(WSNs) has grown dramatically in the past few years, driven by advances in miniaturized hardware, and motivated by a vast array of potential applications. Typical scenarios for large-scale WSNs contain multiple sources and multiple sinks. In WSNs, energy is the most critical resource constraint and the lifetime of a WSNs is dependent on limited battery although there are several promising methods for energy scavenging which need additional circuitry and cost. Besides, performance issues in WSNs play a vital role in some applications as for example message transfer delay for detection systems. Therefore, in this research, we focus on the sink placement problem for minimizing the maximum worst-case delay as well as maximizing the lifetime of a WSN. Di erent sink placement strategies are introduced and their advantages and disadvantages are discussed. A random sink placment (RSP) strategy is used as a lower bound for other strategies. A geographic sink placement (GSP) strategy is proposed as a benchmark sink placement in WSNs whereas an intelligent sink placement (ISP) strategy is introduced as an optimal sink placement. Due to the computational expensiveness of ISP, a Genetic Algorithm-based sink placement (GASP) strategy is discussed and is demonstrated to represent a good approximation for the optimal sink placement. During the performance analysis of the sink placement strategies, the question of a good number of sinks given a certain number of sensor nodes is also discussed. To calculate the maximum worst-case delay, sensor network calculus (introduced in [8]), a recent methodology, is applied in this research work. The core of this research work is to give a framework for sink placement in real-time WSNs applications with two GSP and GASP strategies. The GSP strategies are suited for WSNs in which sensor nodes positions are unknown, whereas the GASP strategies are designated for WSNs when given sensor nodes' positions. Finally, numerical results are obtained using DISCO network calculator [7] for di erent scenarios.