This paper analyzes the characteristics of narrowband power line communication (NB-PLC) channels and assesses their performance when used for advanced metering infrastructure (AMI) communications. This medium has been traditionally considered too hostile. However, the research activities carried out in the last decade have shown that it is a suitable technology for a large number of applications. This work provides a statistical characterization of NB-PLC channels in the CENELEC-A band. The presented results have been obtained from a set of 106 links measured in urban, suburban, and rural scenarios. The study covers the input impedance of the power line network, the channel response and the noise. The analysis of the channel response examines the delay spread, the coherence bandwidth, and the attenuation, while the assessment of the noise considers both its spectral and temporal characteristics. Since low voltage (LV) distribution networks consists of several conductors, they can be simultaneously used to set up multiple-input multiple-output (MIMO) communication links. This paper investigates the correlation between the MIMO streams. The bit rates that can be attained both in the single-input single-output (SISO) and in the MIMO cases are estimated and discussed.

This paper analyzes the suitability of the Middleton canonical class A distribution to model the noise amplitude in narrowband power line communications (NB-PLC). The study has been accomplished using 311 noise registers measured in the CENELEC-A band.Obtained results indicate that this model has a quite limited capability to represent the measured noise, since only 14.47% of the registers have amplitudes drawn from this distribution. Nevertheless, the analysis of these successful cases provides practical information for the utilization of the model. Hence, 56% of these cases can be adequately modeled by truncating the infinite sum in the class A probability density function (PDF) to two terms, which makes it equivalent to the Bernoulli-Gaussian model. It is also shown that the values of the impulsive index and the Gaussian to impulsive power noise ratio are inversely related.

No abstract

No abstract

No abstract