Combinatorial Optimization for Artificial Intelligence Enabled Mobile Network Automation

This chapter discusses combinatorial optimization techniques for enabling intelligent automation in mobile networks. A number of discrete optimization problems pertinent to mobile network automation can be solved effectively using artificial intelligence based combinatorial optimization approaches such as heuristics and metaheuristics. Relevant use-cases include both initial parameter assignment during network roll-out, and continuous optimization of configuration management parameters during network operation and maintenance. We discuss mobile network automation use-cases and motivation for using different heuristics and metaheuristics in designing network optimization algorithms. To this end, we review important metaheuristics from a network optimization perspective, and discuss their applications in different mobile network automation use-cases. As a case study, we discuss greedy heuristics for physical cell identifier (PCI) assignment problem, which is an important use-case relevant to both 4G and 5G networks. The performance of algorithms is compared using a network model based on data from a real LTE mobile network. Results show that greedy heuristics constitute a viable approach for PCI assignment in highly dense networks. We conclude that heuristics and metaheuristics based combinatorial optimization algorithms are highly effective in meeting emerging challenges related to network optimization, thereby enabling intelligent automation in mobile networks.