Self-learning Prioritization of End-User Services to Eliminate Overload Situations

Telecommunication network providers employ various strategies to protect from and to mitigate overload situations caused by signaling storms to minimize end-user service loss. The common approach is the deployment of additional hardware above the engineered capacity combined with resource intensive operational recovery procedures. While signaling storms are relatively rare in its occurrence, they usually have serious consequences - loss of end-user service resulting in negative publicity and business damage. Adaptive overload management emphasizes end-user service as its primary goal in addition to the protection of a network function. The communication dialogs necessary to establish the end-user service are automatically detected and the involved requests are appropriately prioritized. Combining these two processes, the probability of service establishment and its eventual restoration is increased, which contributes to the reduction of overload situation as more end-users can receive its service. Self-learning request prioritization can reduce the time and complexity needed to restore service for all end-users during signaling storms. Through its automatic and self-learning operation it is suited for current and upcoming cloudified and 5G core networks.