Partial Distributed Quantization Aware LMS (PDQA-LMS) algorithm for privacy preserving IoT applications

In this paper, for the first time, we combine the assumption of partial data exchange over adaptive networks with the quantization-aware diffusion least mean square (DQA-LMS) to propose the Partial DQA-LMS (PDQA-LMS) algorithm. This algorithm offers reduced computational and communication complexity by utilizing partial data updates and lower power consumption through the implementation of low-resolution analog-to-digital converters (ADCs) for the internet of things (IoT) applications. Our simulations focus on two key parameters: the number of communicated estimation vector elements in the partial update and the number of quantization bits in the ADC of each node in the adaptive network. The results demonstrate that the number of quantization bits has a more significant impact on the network performance compared to the number of estimation vector elements. Overall, the PDQA-LMS algorithm offers the potential to achieve energy-efficient distributed estimation while maintaining desirable performance characteristics for IoT applications such as privacy preserving.