Hybrid Channel Alloction Scheme Continue...

Cox and Reudink were the first researchers to present published comparisons of different channel allocation schemes. Their comparison was based on simulation of an outdoor vehicular wireless communication system [Cox - 1971, Cox - 1972, Jakes]. The simulation divided a region into a grid of square cells. The movement of vehicles had a two dimensional normal distribution with 0 mean and 30 mph standard deviation in each of the two orthogonal directions. Poisson arrivals were assumed for the rate of calls per vehicle and call durations were assume to have a truncated normal distribution (truncated on the left at zero) with a "mean" 90 seconds (true mean of 103.5 seconds).

Cox and Reudink's study considered uniform and non-uniform distributions of spatial traffic. In the uniform case, all cells had approximately the same call arrival rate while in the non-uniform case, some cells had a significantly higher call arrival rate. With both the uniform and non-uniform spatial distributions, fixed channel allocation schemes were optimally matched so that the cells with the greatest numbers of calls had the greatest number of channels to deal with those calls. In both cases of uniform and non-uniform traffic, results showed that for low blocking probabilities, dynamic channel allocation schemes could handle more calls than fixed channel allocation schemes. More specifically, in the case of uniform traffic, the DCA approach outperformed the FCA approach when the blocking probability was lower than 10%. At a blocking probability of 1%, the DCA approach could handle over 10% more calls than the FCA approach. In the case of non-uniform traffic, the DCA approach outperformed FCA for blocking rates up to 60%. At a blocking rate of 1%, DCA could handle almost 70% more calls per cell than FCA. Cox and Reudink performed another comparison involving dynamic channel reassignment in [Cox - 1973]. In this hybrid procedure, the total number of available channels is broken into two groups: fixed and dynamic channels. When a cell requires a channel, it first searches for an available fixed channel that is preassigned to the cell. If none of the fixed channels are available, a dynamic channel is searched for from the common bank of dynamic channels. If this search is in vain, the call is blocked. When users who were assigned fixed channels end their calls, these freed fixed channels are then assigned to users in the same cell who are currently using dynamic channels. This frees the dynamic channel for future use and ensures that a large number of channels being used are the optimally-spaced, fixed channels. Results from Cox and Reudink's study of dynamic channel reassignment showed that channel use was increased by over 60% compared to fixed channel allocation for a blocking rate of 1%. This result corresponds to uniform offered traffic.