MIXED DISTRIBUTED/CENTRALIZED ROUTING TECHNIQUES BASED ON CLOSED-LOOP FEEDBACK FROM A LEARNING MACHINE TO AVOID DARK ZONES

摘要

In one embodiment, a routing topology of a network including nodes interconnected by communication links is determined, and activity in the network is monitored to determine a normal behavior of the communication links. Weak communication links in the network that deviate from the determined normal behavior are detected, and it is then determined whether the weak communication links are spatially correlated based on the determined topology of the network. In response to the weak communication links being spatially correlated, a region of the network affected by the weak communication links is identified as a dark zone that is to be avoided when routing data packets in the network.

主权项

1. A method, comprising:
determining a routing topology of a network including nodes interconnected by communication links; monitoring activity in the network to determine a normal behavior of the communication links; detecting weak communication links in the network that deviate from the determined normal behavior; determining whether the weak communication links are spatially correlated based on the determined topology of the network; and in response to the weak communication links being spatially correlated, identifying a region of the network affected by the weak communication links as a dark zone that is to be substantially avoided when routing data packets in the network.