Enhancing Rate and Reliability of PLC in LV/MV Smart Grid Networks

Power line communications (PLC) is a promising solution for smart grid communications due to low deployment cost over the existing power line infrastructure. In this dissertation, we propose enhancing the PLC reliability and/or data rate by exploiting the noise and channel statistical properties in time, frequency and spatial domains. For the Narrowband PLC (NB-PLC) in the 3-500 kHz frequency band, a major challenge is the presence of cyclostationary noise whose statistics vary periodically with a period of half the AC cycle. We propose two techniques to mitigate the cyclostationary noise, namely, erasure decoding and noise cancellation. For erasure decoding, we enhance the channel decoding capabilities by feeding the positions of the noise impulses to the channel decoder. Erasure decoding has been investigated for different modes including the Reed-Solomon decoder only, the Viterbi decoder only, and the concatenated decoding modes. Next, we investigate two cyclostationary noise cancellation techniques that offer better performance but at higher complexity, namely, the temporal-region-based and the frequency-shift-filtering-based (FRESH-filtering-based) techniques. The temporal-region-based technique assumes that the cyclostationary noise is stationary over each of the multiple temporal-regions, and employs a per sub-channel linear minimum mean square (LMMSE) estimator in the frequency domain. The FRESH-filtering-based technique aims to filter the cyclostationary noise in the time domain using FRESH filters that utilize time average MMSE (TA-MMSE) estimator. Furthermore, we extend both cyclostationary noise cancellation techniques to exploit the available spatial dimensions, i.e., multiple receive powerline phases. Moreover, to ensure realistic results, we develop novel methods to generate the cyclostationary noise based on FRESH filtering where the filter coefficients are extracted based on field measurements. For the broadband powerline communication (BB-PLC) in the 1.8-250 MHz frequency band, additional diversity can be achieved through simultaneous transmissions over powerline and unlicensed wireless frequency bands, namely, hybrid PLC-wireless system. However, a hybrid PLC-wireless system faces two main challenges. First, in-home BB-PLC systems suffer from impulsive noise (IN). Second, unlicensed wireless transmissions are subject to narrowband interference (NBI) from other in-band wireless communication systems. Therefore, we propose a new sparsity-aware framework to model and mitigate the joint effects of NBI and IN in hybrid PLC-wireless system. In addition, we explore different cases of the NBI and IN including the block sparse NBI/IN and asynchronous NBI cases. For further mitigation performance enhancements, we investigate a Bayesian LMMSE-based approach. Numerical results show superiority of our proposed joint processing of NBI and IN sparsity-based mitigation techniques versus separate processing. Lastly, we investigate the application of multi-input multi-output (MIMO) orthogonal frequency division multiplexing (OFDM) to NB-PLC over medium-voltage (MV) underground networks. We study different MIMO transmission scenarios with different injection configurations utilizing both the cable conductor and sheath phases. Multi-conductor transmission line theory is used to characterize the underground MV NB-PLC channel transfer function. The achievable data rates are evaluated after optimizing the transmit energy allocation across different spatial information streams subject to a power constraint. The achievable data rates for MIMO configurations are shown to be significantly higher compared to singleinput single-output OFDM transmission.