Performance Analysis of Threshold Call Admission Policy for Multi-class Traffic in Low Earth Orbit Mobile Satellite Systems

On Channel Sharing Policies in LEO Mobile Satellite Systems

“…As α 1 continues to increase, the corresponding probabilities of the 1 st service-class calls will increase until another block of 19 channels becomes available to 1 st service-class calls. Such oscillations have not been studied in [15]- [17] and show that attention is needed when dimensioning a system, especially when calls of a serviceclass require much more bandwidth than others. Note that oscillations do not appear in the case of the 2 nd service-class and thus we do not present the corresponding results.…”

“…Contrary to [15], where only simulation results are presented in the case of the TCA policy, or [17] where the set of GB equations should be solved, we propose the mathematical framework for the efficient calculation of all relevant performance measures.…”

“…In the literature, there are various teletraffic loss or queueing models that describe channel sharing policies in LEO-MSS [4]- [17]. Although there are different ways to classify them, e.g., in terms of the channel sharing policy, the call arrival process, the existence of queues or not etc., for presentation purposes, we classify them in two categories: i) single-rate [4]- [13] and ii) multirate [14]- [17] models.…”

“…In the TCA policy, a new service-class k call is not accepted in a cell if the number of in-service new and handover service-class k calls plus the new call exceeds a threshold (different for each serviceclass). In [15], simulation results initially are presented for the TCA policy, while in [17] an analytical Markovian model is proposed that allows the determination of the performance measures by solving the global balance (GB) equations of K-dimensional Markov chains. This task is computationally extremely complex (if not impossible) and time consuming for systems with large capacities and many service-classes, since it requires the solution of a linear system of millions or even billions of GB equations.…”

“…2) we show that: i) the steady state probability distribution in the TCA has a product form solution (PFS), and ii) the channel occupancy distribution can be easily determined with the aid of a convolution algorithm. Compared to [17], where again enumeration and processing of the state-space is required, the proposed algorithm has a low computational complexity of O(KC 2 ). 3) provide a framework for the applicability of the proposed models in LEO SDN/NFV satellite networks.…”