Effect of Transformation in Compressed Sensing of Smart Grid Data

Control and monitoring of the Smart Grid is facilitated by two way communication of information between agents in the grid eg., smart meters. However, continued rapid deployment of smart meters may lead to data avalanche in terms of volume and velocity, resulting in elevated communication traffic. Compressive sensing is a data compression technique that accounts for sparsity of electricity consumption pattern in a transformation basis and achieves sub-Nyquist compression. As the power consumption data has varying sparseness level, the choice of transformation basis plays a significant role in influencing the compression performance (compression level and reconstruction error). To the best of our knowledge, most of the studies related to compression of power consumption signal assume Wavelets or Discrete Cosine Transform to be the de facto transformation basis. In this regard, we show a comparative study showing the compression performance with different transformations i.e., Discrete Cosine Transform, Haar, Hadamard, Hankel and Toeplitz transformations. We present the results on three publicly available data sets at different sampling rates and outline key findings of the study.