Multi-Task Mapping and Resource Allocation Mechanism in Software Defined Sensor Networks

Traditional wireless sensor network (WSN) is deployed based on specific task lacking of flexibility and ability to process and compute multi-task in parallel, which usually leads to deploy redundant WSNs for diversified Internet of things (IoT) requirements. In order to solve the problem, a software defined wireless sensor network (SDWSN) resource scheduling and mapping mechanism is proposed for multi-task scenarios in WSNs. In particular, SDWSN is introduced to support diversified IoT applications on a same WSN by abstracting the physical resources into logical ones. In order to fully explore the resources of WSN, a network virtual layer is set between data layer and control layer of the SDWSN through FlowVisor, which can enable a sensor node in SDWSN to process and compute multi-task in parallel. In addition, a dynamic alliance is establish for each task through the non-linear weight discrete particle swarm (NWDPSO) algorithm to complete resource mapping from logical network to physical network. The results show that the proposed multi-task resource scheduling and mapping mechanism can not only improve resource utilization and load balancing, but also reduces network energy consumption and task completion time.