An important objective of surveillance sensor networks is to effectively monitor the environment, and detect, localize, and classify targets of interest. The optimal sensor placement enables us to minimize manpower and time, to acquire accurate information on target situation and movement, and to rapidly change tactics in the dynamic field. Most of previous researches regarding the sensor deployment have been conducted without considering practical input factors. Thus in this paper, we apply more real-world input factors such as sensor capabilities, terrain features, target identification, and direction of target movements to the sensor placement problem. We propose a novel and efficient hybrid steady-state genetic algorithm giving low computational overhead as well as optimal sensor placement for enhancing surveillance capability to monitor and locate target vehicles. The proposed algorithm introduces new two-dimensional geographic crossover and mutation. By using a new simulator adopting the proposed genetic algorithm developed in this paper, we demonstrate successful applications to the wireless real-world surveillance sensor placement problem giving very high detection and classification rates, 97.5% and 87.4%, respectively.

In a wireless sensor network (WSN), a large number of sensor nodes are deployed over a wide area and multi-hop communications are required between the nodes. Managing numerous sensor nodes is a very complicated task, especially when the energy issue is involved. Even though a number of ad-hoc management and network structuring approaches for WSNs have been proposed, a management framework covering the entire network management infrastructure from the messaging protocol to the network structuring algorithm has not yet been proposed. In this paper we introduce a management framework for WSNs called SNOWMAN (SeNsOr netWork MANagement) framework. It employs the policy-based management approach for letting the sensor nodes autonomously organize and manage themselves. Moreover, a new light-weight policy distribution protocol called TinyCOPS-PR and policy information base (PIB) are also developed. To facilitate scalable and localized management of sensor networks, the proposed SNOWMAN constructs a 3-tier hierarchy of regions, clusters, and sensor nodes. The effectiveness of the proposed framework is validated through actual implementation and simulation using ns-2. The simulation results reveal that the proposed framework allows smaller energy consumption for network management and longer network lifetime than the existing schemes such as LEACH and LEACH-C for practical size networks.