Stability preserving transformations: packet routing networks with edge capacities and speeds

In the context of an adversarial input model, we consider the effect on stability results when edges in packet routing networks can have capacities and speeds/slowdowns. In traditional packet routing networks, every edge is considered to have the same unit capacity and unit speed. We consider both static modifications (i.e. where the capacity or speed of an edge is fixed) and dynamic modifications where either the capacity or the speed of an edge can be dynamically changing over time. Amongst our results, we show that the universal stability of LIS is not preserved when either the capacity or the speed is changing dynamically whereas many other common scheduling protocols do maintain their universal stability. In terms of universal stability of networks, stability is preserved for dynamically changing capacities and speeds. The situation for static modifications, is not as clear but we are able to show that (in contrast to the dynamic case) that any “well defined” universally stable scheduling rule maintains its universality under static capacities, and common scheduling rules also maintain their universal stability under static speeds.

[1]  Baruch Awerbuch,et al.  Universal-stability results and performance bounds for greedy contention-resolution protocols , 2001, JACM.

[2]  Rafail Ostrovsky,et al.  Adaptive packet routing for bursty adversarial traffic , 1998, STOC '98.

[3]  Baruch Awerbuch,et al.  Universal stability results for greedy contention-resolution protocols , 1996, Proceedings of 37th Conference on Foundations of Computer Science.

[4]  Ashish Goel Stability of networks and protocols in the adversarial queueing model for packet routing , 1999, SODA '99.

[5]  Panayiotis Tsaparas,et al.  Stability in adversarial queueing theory , 1997 .

[6]  David Gamarnik,et al.  Stability of adaptive and non-adaptive packet routing policies in adversarial queueing networks , 1999, STOC '99.

[7]  Rajeev Motwani,et al.  Randomized Algorithms , 1995, SIGA.

[8]  Matthew Andrews,et al.  Instability of FIFO in session-oriented networks , 2000, SODA '00.

[9]  Lisa Zhang,et al.  The effects of temporary sessions on network performance , 2000, SODA '00.

[10]  Gary L. Miller,et al.  Proceedings of the twenty-eighth annual ACM symposium on Theory of computing , 1996, STOC 1996.

[11]  David Gamarnik Stability of adversarial queues via fluid models , 1998, Proceedings 39th Annual Symposium on Foundations of Computer Science (Cat. No.98CB36280).

[12]  Allan Borodin,et al.  Adversarial queueing theory , 1996, STOC '96.

[13]  M. del Carme Álvarez Faura,et al.  A Characterization of universal stability for directed graphs in the adversarial queueing model , 2002 .