Frequency Planning For PHS Network: Techniques And Applications

In this paper, the frequency assignment problem of PHS(Personal Handy-phone System) is discussed and solved. The control stations (CS) of PHS network use DCA (Dynamic Channel Assignment) algorithm to choose its frequency. This frequency assignment method works poorly when the load network is high. Thus we propose to confine the candidate frequencies scanned by DCA to improve the performance of PHS network as well as the feeling of customers.The DCA strategy of PHS system has a fundmental flaw: the wireless environment of a personal station (PS) is not always the same with that of the CS to which it communicates with. When using DCA algorithm, a CS continuously scans all its frequency span according to a given rule, monitoring the noise level on every carrier scanned. When a PS sends in a TCH link establishment request, the control station chooses the carrier which has the lowest noise level for this request. In a practical PHS network, however, this strategy often fails because of the serious asymmetry between PS' s vision and CS' s vision. This paper discussed four factors leading such vision asymmetry: the shelter of buildings, the attenuation of signal, the interaction between directional-antenna CS' s and omnidirectional-antenna CS' s, and the interaction between high CS' s and low CS' s. Because of the vision asymmetry, there is a great probability that a PS would reject the assigned carrier and request TCH link establishment again. As a result, the performance indexes, especially TCH link establishment success rate and TCH link establishment re-request rate, suffer greatly.In an effort to solve this problem, we suggest to perform frequency planning to PHS networks. It is done by confining the candidate frequencies scanned by the DCA algorithm of every CS. We suggest to divide the CS' s into two group: high CS' s and low CS' s first and assign different frequency spans to them, which can alleviate the vision asymmetry arising from the interaction between high CS' s and low CS' s. Then different frequency groups are assigned to each of the two CS groups in order to scattering the CS' s candidate frequencies. Here, the criterion for dividing high CS' s and low CS' s is not CS' s abstractheights, but their relative heights, which is a concept proposed by us to represent a CS' s height, vision and the area influenced by its signal. In addition, three different algorithms: heuristic algorithm, greedy algorithm and Simulated Annealing algorithm are applied to scattering CS' s candidate frequencies. In a practical PHS network, the CS count is usually very large with new CS' s added to the network from time to time, and the channel number of every CS also varies. Therefore, although these three algorithms are based on earlier presented frequency assignment algorithms, they have a lot of important modifications to take the complexity of practical network into consideration.Finally, the proposed frequency planning method is applied to a practical PHS network on operation to manifest its effectiveness. Different tests are carried out and their results are reported and evaluated. The result shows that the proposed frequency planning method improved the performance of the PHS network greatly.