Energy consumption of algorithms for data retrieval from indexed parallel broadcast channels

摘要

Broadcasting over parallel channels has been shown to be an effective method for disseminating public data to mobile devices. Using recent techniques, equipped devices can leverage energy-saving modes to reduce the amount of energy consumed by wireless communication hardware. However, the use of parallel channels leads to challenges accessing concurrently broadcast data items and switching between channels both of which can incur additional energy costs and increase response time. To reduce these costs, the retrieval of data items can be scheduled in an order that reduces the required number of channel switches and repeated passes over a broadcast. Existing scheduling techniques, including Parallel Object Scan (POS) and Serial Empty Scan (SES) are able to produce optimal schedules through a broadcast on parallel channels. The energy consumed by the generated schedules has been studied previously. In this paper, we investigate the energy consumed by the execution of the POS and SES scheduling algorithms. The algorithms were executed in an emulated Android environment, where instrumentation tools were used to monitor the instructions executed by the virtualized CPU. Then, the total energy consumption for each algorithm was calculated using average energy costs per instruction.