Abstract
Wireless Sensor Networks (WSNs) support a variety of data collection scenarios and have profound effects on both military and civil applications, such as environmental monitoring, traffic surveillance and tactical military monitoring. Design of efficient data collection algorithms is important yet still challenging due to the distinguished characteristics of WSNs: (i) The large number of sensor nodes may cause severe unbalanced traffic through the network due to the concentration of data traffic towards the sinks and the intersection of multihop routes. (ii) Sensor nodes are limited in power, computational capability and storage capacity, which requires careful resource management using energy efficient schemes. (iii) WSNs are typically application-specific, and the design requirements of networks change with different applications. This thesis presents the following three contributions to the literature of efficient data collection in WSNs: First, we proposed a unified solution for gateway and in-network traffic load balancing in multihop data collection scenarios. We combined multiple path metrics (path residual bandwidth, end-to-end delay and path reliability) and gateway conditions (gateway utilization) in a unified path quality metric. The strategy is to probabilistically choose alternative path and adaptively modify the path switch probability based on the independent decisions made by the sensor nodes. Second, we formulated the delay aware energy efficient data collection with mobile sink and virtual multiple-input multiple-output (VMIMO) technique problem and proposed a weighted revenue based algorithm to approximate the optimal solution. The aim is to achieve full utilization of VMIMO technique to minimize the network energy consumption with consideration of bounded sink moving time. In order to explore the trade-off between overall network consumption and data collection latency, we combined the VMIMO utilization, and sink moving tour length into a weighted metric. Third, we established an minimization model for the total data collection latency in multihop data collection scenarios with bounded hop distance and limited buffer storage. To approximate the optimal solution, we developed a multihop weighted revenue algorithm. The strategy is to jointly consider data uploading time and sink moving time to optimize the total data collection time. In order to increase the time saving due to concurrent data uploading, we balanced the number of associated nodes of the compatible sensors.