Characterizing node availability and connectivity of a hybrid optical/RF wireless network

The performance of wireless nodes configured with hybrid free space optical (FSO) and radio-frequency (RF) links is analyzed in this thesis. The main objective was to determine the number of hybrid links that were required at each node of the wireless network, so that specified availability metrics would be met. To model the random occurrence of link outages each FSO and RF link configured at network terminal was represented by a two state on-off Markov chain with independent state transitions. Using the duty cycle of the hybrid link as the characterizing parameter, it was shown that atleast four hybrid link pairs would be required to achieve 99% availability of atleast one RF link capacity for duty cycles greater than 50%. To ensure 99% availability of atleast one FSO link capacity, the FSO channel must exhibit duty cycles greater than 0.7. Using two state approximation of a multi-terminal node, the blocking, delay and throughput analysis for multihop transmission has also been presented. Analytical expressions derived for the buffer occupancy show its explicit dependence of the average off duration of the hybrid channel. Both first and second moments of buffer occupancy tend to be higher than that found for Bernoulli or deterministic channels. The blocking analysis showed that by providing a waiting time upto 160 milliseconds and with four hybrid link pairs configured at each node, the blocking probabilities can be limited to the order of 10-4 packets. Using a linear programming approach the access and backbone network connectivity was examined for the case of on-off channels. It was shown that the average of two hop counts required when all links were on increases to an average of four hop counts with links turned off independently. The suite of analysis and optimization properties produced in this thesis can be relatively used to estimate availability of node, their blocking and delay performance and the performance at a network level.