A recommender-system for telecommunications network management actions

Research and products in telecoms network management have long been focused on the automation of processes to keep complex managed networks in an operational state, while being profitable to operate for the network operator. Due to the ever increasing scale and complexity of the problem domain, coupled with specific constraints (legal, regulatory, technological change), a fully automated management approach is virtually impossible. It remains the responsibility of human Network Operations Centre (NOC) operators to oversee and manage the running of the network, where their main task is to respond to huge numbers of messages, errors, warnings and faults constantly flowing from the managed network. In the past, network operators either employed expert systems or became very dependent on expert knowledge of their operational staff. Considering the current explosion in size and complexity of managed telecoms network, it is widely understood and accepted that current manual and semiautomated approaches cannot scale. In this paper we investigate the applicability of recommender systems as an approach to assist NOC operators to correctly respond to indications of incidents in the network they are actively managing.