Space-Optimal Packet Routing on Trees

We consider packet forwarding on a tree with all packets destined for the root, assuming each link may forward at most$c \geq 1$ packets each time step. We use the Adversarial Queuing Theory injection model, where a$(\rho,\ \sigma)$-adversary may inject at most$\sigma+\rho\cdot t$ packets into the network at arbitrary locations during any time interval of length t. The goal is to find a forwarding protocol that minimizes the maximal butter space required to avoid overflows against a$(\rho,\ \sigma)$-adversary with$\rho\leq c.$ We consider protocols from the locality viewpoint. A protocol is called d-local if the actions of a node depend only on the current state of nodes at distance at most d. A D-local protocol, where D is the network diameter, is called centralized. It is known that buffers of size$\Theta(\sigma+\rho)$ are necessary and sufficient for centralized protocols. The butter requirement of$O(1)$-local protocols was recently proved to be$\Theta(\rho\log D+\sigma)$. In this paper, for any$d \geq 2$, we describe a d-local algorithm whose butter space requirement is$O \left( \left\lceil \frac { \log D } { d } \right\rceil \rho + \sigma \right)$. This result is tight, up to constant factors. In particular, it implies that$O(\log D)$ locality is sufficient to achieve the best worst-case performance possible even for centralized algorithms. We also give evidence suggesting that the butter requirement of a local algorithm designed for trees is good also when the routes do not constitute a single-destination tree.

[1]  Jean C. Walrand,et al.  Achieving 100% throughput in an input-queued switch , 1996, Proceedings of IEEE INFOCOM '96. Conference on Computer Communications.

[2]  Yossi Azar,et al.  Packet Routing and Information Gathering in Lines, Rings and Trees , 2005, ESA.

[3]  Michael H. Goldwasser A survey of buffer management policies for packet switches , 2010, SIGA.

[4]  Boaz Patt-Shamir,et al.  The Space Requirement of Local Forwarding on Acyclic Networks , 2017, PODC.

[5]  Rene L. Cruz,et al.  A calculus for network delay, Part I: Network elements in isolation , 1991, IEEE Trans. Inf. Theory.

[6]  J Gettys,et al.  Bufferbloat: Dark Buffers in the Internet , 2011, IEEE Internet Computing.

[7]  Allan Borodin,et al.  Adversarial queuing theory , 2001, JACM.

[8]  Adi Rosén,et al.  Rate vs. buffer size: greedy information gathering on the line , 2007, SPAA '07.

[9]  Kirk Pruhs,et al.  Packet Forwarding Algorithms in a Line Network , 2014, LATIN.

[10]  Boaz Patt-Shamir,et al.  Buffer Size for Routing Limited-Rate Adversarial Traffic , 2016, DISC.

[11]  Christian Scheideler,et al.  Information gathering in adversarial systems: lines and cycles , 2003, SPAA '03.

[12]  Adi Ros Tight Bounds for the Performance of Longest-in-System on DAGs , 2002 .

[13]  Lata Narayanan,et al.  Optimal Local Buffer Management for Information Gathering with Adversarial Traffic , 2017, SPAA.

[14]  Adi Ros A Note on Models for Non-Probabilistic Analysis of Packet-Switching Networks , 2002 .