Signal-Processing-Aided Distributed Compression in Virtual MIMO-Based Wireless Sensor Networks

An adaptive signal-processing-aided distributed source coding scheme for virtual multiple-input-multiple-output communication-based wireless sensor networks (WSNs) is proposed. A computationally inexpensive distributed compression scheme that exploits the spatiotemporal correlations of sensor data is implemented with the aid of a recursive least squares (RLS)-based adaptive correlation tracking algorithm. The tracked correlation is used to compute side information that assists in distributed source compression. The proposed virtual space-time block coding and RLS-based compression side information are shown to improve energy efficiency at distributed nodes compared to previously proposed schemes with single-input-single-output communication. A semi-analytical approach is developed for energy efficiency analysis over different channel conditions and transmission distances. The energy efficiency performance of the proposed design is evaluated on real WSN data. The results show that the proposed integrated system outperforms conventional designs beyond certain transmission distance thresholds and leads to lower decoding errors, which makes it a good candidate for energy-aware WSNs.