Energy efficient medium access control protocol for data collection in wireless sensor network: A Q-learning approach

Abstract

In emerging business markets, the data is becoming new gold for the industries. There are several places such as marketing, manufacturing, and analysis of future trends becoming data-intensive to achieve growth. The tiny sensor placements in the field, cities, industry, buildings and sea are helping to collect data and process it either for information retrieval or decision making. The periodic scheduling of radio transceivers assists in accomplishing efficient energy utilization in sensors often uses Time-division multiple access (TDMA) protocols for node scheduling. Our objective is to reduce the amount of time; the receiver node is in the wake-up state. Besides, the slot which potentially not being used for an extended period can utilize by other nodes. The efficient utilization of slots can help to achieve low power duty cycles with low latency. To accomplish that, we propose the emulation of the classroom learning environment with the Q-learning grading for node scheduling. We proposed an analytical mapping of WSNs to classroom learning. The initial benchmark of the performance has compared to the IEEE.802.11 (TDMA-scheduling). Further, the TDMA driven protocols such as Z-MAC, learning-driven protocols (E-MAC or aloha-Q), i-Queue are compared to evaluate the parameters such as energy consumption, throughput, and latency.