Lightweight energy-efficient framework for sensor real-time communications

摘要

Sensor node communications are usually the most energy consuming among its activities. Therefore, reducing those communications necessarily preserves the node limited energy. One successful approach to this end is to reduce both the number of transmissions made by the node and the payload of each transmission. Numerous proposals have recently been made to achieve this reduction by approximating the node data at the sink via forecasting. However, many of these proposals are plagued with one or more of three problems: intensive computations, excessive delays and backtracking. In this study, the authors introduce the lightweight energy-efficient real-time (LEERT) framework, which avoids all three problems while saving considerably on energy consumption. The savings are achieved by both reducing immensely the number of transmissions and reducing the payload of each transmission to only one element. This is actually the minimum any framework would hope for. LEERT is also extremely light, performing only four primitive operations per measurement on average. Its storage requirements are even lighter - only one past measurement needs to be retained in memory. LEERT has passed extensive validation tests using real-world data. Its performance has been evaluated against two recent energy-efficient frameworks and found to have an astounding edge.