Abstract
A novel flexible channel allocation scheme is proposed which is applicable to medium altitude, high diversity satellite personal communication systems with an interconnected ground control segment. The scheme is built upon the establishment of several temporal master-slave network hierarchies in distinct areas of the Earth, whereby the control of radio resources in each area is granted to a single network entity, the master, located at one of the network's fixed earth station sites. The master station is entrusted with the radio resource management functionality equivalent to a base station controller in GSM, whereas the slaves, which consist of all FES sites which control satellites that have a current connectivity with the master's area, share the functionality of a GSM base transceiver station with the satellites under their control. The master maintains a table of all channel allocations within its control area in real-time, its objective is to maximise the number of channels which can be provided in its area, in response to channel requests by users, subject to known interference criteria. It can achieve this by optimising the existing and future channel allocations through the use of predictive satellite coverage, spotbeam gain models and genetic algorithms. Radio resource related signalling between master and slave stations is performed through a dedicated terrestrial fibre-optic network which interconnects the FES sites, based upon GSM interface specifications which are adapted to account for the increased propagation delays and signalling load associated with the regional network. The potential capacity gain which is available through a centralised optimisation approach over a fixed channel allocation scheme is determined through simulation. The instantaneous capacity of each scheme is highly dependent on the traffic distribution and the positions of the satellites, which causes an analytical approach to be too restrictive or simplistic. The capacity gain estimation is performed over a time period for which several samples are taken and a powerful adapted genetic algorithm is applied to the NP complete problem to determine the maximum capacity. The flexible dynamic system, entitled the hybrid-centralised DCA scheme, is shown to have a raw capacity potential in excess of the FCA scheme under the traffic distribution scenario adopted.