A channel assignment problem is to develop efficient channel assignment policies that can maximize the number of calls served in cellular mobile systems. This problem is one of the most important problems in the design and operation of cellular mobile systems. In the future, the number of subscribers will increase more rapidly and hence the number of cells will also increase more rapidly. Thus this problem will be much more important.
Channel assignment policies can be roughly classified into two categories: fixed channel assignment (FCA) and dynamic channel assignment (DCA). FCA is to assign a fixed channel to a call appearing in each cell, where the fixed channel is one of channels allocated to the cell in advance. In FCA, it counts how channels are allocated to each cell in advance. On the other hand, DCA is to assign a fixed or dynamic channel to a call, where the dynamic channel is one of channels allocated to common pools and used in several cells. In DCA, it is interesting how dynamic channels are assigned to calls in each cell. In this thesis, we have dealt with a minimum span problem for FCA, and a hybrid channel assignment (HCA) problem and a borrowing channel assignment (BCA) problem for DCA.
MSP is to minimize the span, i.e., the number of channels which are required to satisfy a given grade of service for a cellular system, subject to the channel separation constraints. To solve the problem, we have suggested how to reduce a considered system to a smaller system by using a clustering concept, and how to partition the reduced system into several parts, and then a heuristic algorithm based on these two methods. A basic idea of solving a partitioned problem is to maximize the separation between channels assigned to any two nearest cells. We have compared the computational results of our algorithm with those of existing algorithms.
In HCA, each cell has more than two types of channel sets, where one type of channel set consists of fixed channels and anther type of channel set consists of dynamic channels in a common pool. Thus we are interested in which channel set is used when a call appears in each cell. Using (a channel in) a channel set causes increases the future call reject rate, i.e., the probability that each of the future calls appearing in the channel set will be rejected. Under some assumptions, we have suggested how to compute the future call reject rates, and shown that assigning a new call appearing in each cell to a channel in a channel set with the lowest probability minimizes the number of rejected calls. With numerical examples, we have compared the blocking probability of our HCA with those of FCA and maximum packing for DCA.
BCA is to assign a fixed channel to a call appearing in each cell when the fixed channel is free and to borrow a channel from a neighboring cell when no fixed channel is free. Thus we are interested in which channel is borrowed from a neighboring cell. If a cell borrows a channel from a neighboring cell, the borrowing cell keeps the channel from being using in some cells within the interference of the channel. The borrowing cell becomes a locking cell for the channel in each of some cells. Channel borrowing gives a bad impact to the channel availability, i.e., the ratio of the available channels to the total fixed channels, of each affected cells. We have estimated the number of the available channels in each cell as the summation of future channel availability of each fixed channel in the cell. Our proposed channel borrowing policy is to borrow a lowest numbered channel which gives the minimum impact. Channel rearrangement can produce a free channel even when no channel is free for a new call appearing in a cell. We have added a channel rearrangement procedure which can use a channel having only one locking cell for a new call and use a borrowed channel for the corresponding call. Simulation results have shown that our policies reduce remarkedly blocking probabilities and also the number of rearrangements over existing policies.
