Compressive sensing (CS) is an effective strategy for data collection and maintaining energy consumption balance in wireless sensor networks (WSN). CS usually exploits the space-time correlations of signal information and the compressed data acquired with this property from the remote field. This research proposes an efficient data-gathering method based on CS, which performs sequential sampling with the progressive reconstruction of sensor data through joint data dependencies. The proposed enhanced energy efficient distributed compressive sensing (EEEDCS) utilizes ℓ2-regularization with iterative re-weighted ℓ1-minimization(IRW-ℓ1) for an estimate of the current signal measurements and regularly updates the previous signal measurements and provides better reconstruction accuracy. Extensive experiments were performed to analyze the proposed method with different network topologies and correlation ranges. Experimental simulations are performed with varying topologies of the network as 49 nodes, 64 nodes, 81 nodes, and 100 nodes. Under 100 node topology, the proposed method saves energy by 8.95%, 14.65%, 20.71%, 22.93%, 25.98%, and 29.24% compared with the baseline models at 40% sampling rate. Also, the proposed EEEDCS method was evaluated with the Pacific Sea Surface Temperature dataset. The Proposed EEEDCS saves energy by 8.76%, 13.97%, 18.18%, 23.90%, 33.14%, and 39.57% compared with the baseline models at a 40% sampling rate. From the results, the proposed model accurately reconstructs the signal samples, and shows its effectiveness over the baseline models considered for comparison, consumes less energy for data collection, and extends the lifetime of sensors and WSNs.
Read full abstract