Partial discharge impulsive noise in 735 kV electricity substations and its impacts on 2.4 GHz ZigBee communications

Installation of wireless sensor networks in power substations is becoming unavoidable for developing smart electrical network. High-voltage equipment in a power substation may produce impulsive noise that happens to be much more troublesome for wireless communications than conventional white noise. Indeed, the choice of physical layer for communication network is not easy. It fits to evaluate attentively the advantages and downsides bound to the different technologies of communication. Before deploying the network, electromagnetic environment has to be characterized and tests have to be performed in order to estimate communication performances. This paper presents an investigation of ZigBee (IEEE 802.15.4/2.4 GHz) narrowband system performance in high voltage substations. We use recorded data from the measurement campaign to identify the models. Substation noise obtained is modeled as a Middleton Class A process and a recent model (Au). By using several statistical tests, we evaluated the two models. First, the results show that Middleton Class A is not accurate for modeling partial discharges at 735 kV electricity substations. Au model is more accurate to model impulsive noise in high voltage substation compare to Middleton Class A and Gaussian models. In addition, the BER performances of the ZigBee 2.4 GHz PHY layer are studied in a substation environment. The results show that the impulsive noise influence is close to a Gaussian noise or a Rayleigh noise according to the SNR.