Load-balanced Data Gatering Algorithms In Wireless Sensor Network

As a cutting edge research area in the automatic control and artificial intelligence, Wireless Sensor Networks (WSN) is a fundamental of the network infrastructure for Pervasive Computing environments, with very wide application prospects.The data gathering application is one of the widespread application forms of WSN, characterized with the large continuous data flow ability and "many-to-one" feature. In data gathering process, the load imbalance of each data stream branch would cause the fast energy exhaustion in overburdened nodes which would leads to the death prematurely. The load balancing technology to alleviate network congestion, which improves network service quality and resource utilization, is very effective and widely applied to the Internet network. However the study of load balancing for the wireless sensor networks is still in its early stages.To improve network service quality and extend the network lifetime, the wireless sensor network load balancing algorithm for data gathering is studied with the following scenarios, respectively:â‘ The stationary Sink node and sensor nodes with the same data generating rate(Homogeneous Network);â‘¡The stationary Sink node and sensor nodes with the different data generating rate(Homogeneous Network);â‘¢The moving Sink node (mobile user network).The main research achievements are listed in the following:1)A dynamic overlapping backoff window method is proposed.In response to the scenariosâ‘ andâ‘¡, the flooding mechanism is approached to build the data gathering tree or layer discovery. Considering the traditional back-off mechanisms in MAC layer would result in severe collisions and route circumambulating in flooding, the Dynamic Overlapping Back-off Window (DOBW) algorithm was presented. According to current rate of neighbors, sensor nodes adjust their back-off windows automatically to reduce message collision and optimize the topology. Comparing with 802.11 and 802.15.4, the simulations shown that DOBW algorithm proposed in this thesis can significantly reduce the impact of flooding and optimize the topology of the data gathering tree.2)A load-balanced data gathering tree generation algorithm is proposed.Though, the DOBW algorithm could construct a certain degree of load-balanced tree, the requirements of load balance is still un-achieved. Therefore, the DOBW algorithm is commonly used in the layer discovery process to avoid the message collision. The Load-balanced Data Gathering Tree based on SPT (LDGT-SPT) to establish a shortest load-balanced path tree was presented for scenarioâ‘ . The LDGT-SPT is consisted of the neighbor discovery, the layer discovery, the priority principle of minimal degree and the traffic balance strategy. Simulation results show that LDGT-SPT algorithm can significantly improve the performance of the network compared with SLBT with the same network lifetime.3) An ACO-based dynamic load balancing algorithm for data collection is proposed.In the scenarioâ‘¡, due to different generating rate of sensor node data, the load-balancing data gathering tree construction methods is un-available, thus the dynamic load balancing approach is adapted. In this thesis, the Load-balanced Data Gathering algorithm based on Ant Colony Optimization (LDG-ACO) was presented. In order to achieve dynamic load balancing, LDG-ACO algorithm classifies ants to have different functions, taking the node load information as inspiration factor, making ants with the load-sensing function and working according to the principle of lower pheromone higher probability. Simulation results show that compared with the ACO, SLBT and DLBT algorithm, LDG-ACO algorithm presented in this thesis has increased markedly in network performance and network service life.4)A dynamic load balancing algorithm supporting mobile Sink for data collection is proposed.In response to the scenarioâ‘¢, the Mobile Sink led to frequent route changes and link interrupt problem, a mobile Sink dynamic load-balancing algorithm for data collection (LDG-MS) is presented to solve these problems. Referring from the swarm intelligence algorithm, LDG-MS defines two simple rules to describe the data forwarding. The problem how to choose next hop is described as a multi-objective programming. Furtherly, the evaluation method of weighting distance is used in the multi-objective programming. To solve link-break problem, a method of the power control for the Sink beacon messages was proposed. Simulation results show that, compared with SINK_CLAIM, SLM algorithm, LDG-MS algorithm presented in this thesis in network performance and network lifetime has increased profoundly.