Incorporating Imprecise Computations in Scheduling of Aggregated Data within Wireless Sensor Networks

This paper proposes a model to be utilized during the schedule of non-preemptive, imprecise computing tasks within wireless sensor networks. We have incorporated the three tasks of a sensor as sensing, processing and transmission into a task flow graph. In real-time systems, the overloaded sensors may produce approximate results by terminating the execution before completion due to the deadline requirements. This variation in the execution timings of tasks within sensor results in imprecise computations. We have scheduled the data with imprecise task timings based on available energy, aggregated at the gateways using as soon as possible (ASAP) and as late as possible (ALAP) algorithms subject to concurrency in transmissions. We have analyzed the performance of each algorithm in scheduling of 25 sensors by varying the availability of gateways from 1 to 5. The computational results have provided excellent bounds on the number of gateways used for scheduling the transmission and also showed a reduction in the total execution time of the sensor nodes.