MOSKOU: A Heuristic for Data Aggregator Positioning in Smart Grids

Abstract

Smart grids represent a revolution in efficient energy distribution, requiring a series of new devices that communicate with each other. Thus, in addition to energy, smart grids also induce an information flow. The advanced metering infrastructure (AMI) is a middle-range communication network which provides communication between clients’ smart meters and data aggregation points (DAPs). This paper focuses on determining the minimum number of DAPs and their positions such that every smart meter is covered and able to communicate through good quality wireless links. In our proposal, the AMI planning is seen as an instance of the set covering problem. This instance is built by determining all good quality wireless links between smart meters and potential DAP installation sites. The proposal is further extended to consider the possibility of multi-hop communication and to find solutions with a given minimum coverage redundancy for each meter. We solve the problem by applying a heuristic, named memory oriented split using ${K}$ -means with post-optimization unification (MOSKOU), which splits the original problem into smaller ones that are optimally solved. The split is based on the ${K}$ -means clustering algorithm. Results show that MOSKOU is capable of finding near-optimal solutions for huge instances while greatly reducing the amount of RAM memory needed and execution time.