Guaranteed maximal matching for input buffered crossbar switches
While many scheduling algorithms have been proposed so far for input buffered crossbar switches, the challenge still remains to develop low complexity scheduling algorithms. These algorithms should be easy to implement and provide higher throughput and better utilization of switch space under both uniform and non-uniform traffic arrivals. In this paper, we propose modifications for two established algorithms: iSLIP and Maximal Weighted Matching (MWM) in order to reduce the wastage of time slots and thus, to provide better utilization of the switch bandwidth. We, then, propose one new algorithm named as Guaranteed Maximal Matching (GMM). The proposed GMM algorithm ensures to provide perfect matching for uniform traffic arrival and guarantees to extract all possible permutation matrices from the traffic demand matrix under the scenario of non-uniform traffic arrival. We have verified our proposed algorithms for i.i.d traffic demand matrices for NxN crossbar switches of different sizes. Modified MWM and GMM achieve high throughput and perfect matching for uniform traffic while, for non-uniform traffic, GMM outperforms the other algorithms. It performs better for larger switch configurations, offers simpler implementation (maximum complexity in O(N5)) and takes lower computation time.
|Keywords||Crossbar switches, Guaranteed maximal matching, iSLIP, MWM, Perfect matching, Scheduling|
|Conference||4th Annual Communication Networks and Services Research Conference, CNSR 2006|
Al Sayeed, C.A. (Choudhury A.), & Matrawy, A. (2006). Guaranteed maximal matching for input buffered crossbar switches. Presented at the 4th Annual Communication Networks and Services Research Conference, CNSR 2006. doi:10.1109/CNSR.2006.31