Self-organizing networks, or SON, are mobile networks that automate tasks operators traditionally did manually: planning new cells, configuring them, tuning them as conditions change, and recovering from faults. The concept was standardised by the 3GPP in technical specification TS 32.500, “Self-Organizing Networks (SON); Concepts and requirements,” first issued for LTE in Release 8 and carried forward into later releases. The driver was cost: manually generating cell neighbour relations and optimising thousands of base stations is slow and expensive, so automating it reduces operating expense.
SON is usually described in three families of functions. Self-configuration sets up new network elements automatically, including features like Automatic Neighbour Relations that discover and link cells without an engineer entering each relationship. Self-optimization continuously tunes the live network for goals such as mobility robustness, load balancing, and reducing interference between cells. Self-healing detects and works around faults. The 3GPP standard also defines where the algorithms run - centralised at the management layer, distributed in the base stations, or hybrid.
As networks grew through 4G and into 5G, the volume of cells and parameters became too large to manage by hand, and SON increasingly drew on machine learning. For a general reader, SON is a long-running, large-scale example of AI and automation quietly running critical infrastructure that most people never see.