Topology Reconfiguration with Successive Approximations

Abstract

Recent technologies have enabled the formation of point-to-point directional wireless networks that are capable of dynamic changes in the network topology. The process of changing this topology in response to changes in available link capacities and load demands of various nodes is called topology control. One example of the type of communication network studied in this context is a Free Space Optical (FSO) network.

Topology control consists of computing new topologies to dynamically optimize the network under changing traffic conditions and then carrying out the reconfiguration process to achieve the target topology. This thesis considers the process of topology reconfiguration and use the packet drops that happen during this process as a cost metric for this process. It is shown that by implementing the topology reconfiguration as a series of smaller steps (successive approximation), the number of packets that are dropped during the reconfiguration are reduced. Using this knowledge, the topology computation algorithm can be refined to also minimize the reconfiguration cost along with the typical objective of minimizing congestion.