Deterministic network exploration by a single agent with Byzantine tokens

A mobile agent has to explore an unknown network with unlabeled nodes: it must visit all nodes by walking along the links of the network, and eventually stop. If no upper bound on the size of the network is known and nodes of the network cannot be marked, then this exploration task cannot be accomplished for arbitrary networks by a deterministic terminating algorithm. On the other hand, it is feasible, if there is one unmovable token at the starting node of the agent. We investigate the exploration problem in arbitrary networks in the presence of identical unmovable tokens, some of which are Byzantine. A Byzantine token can be visible or invisible to the agent whenever the latter visits the node where the token is located, and visibility is decided by the adversary at each visit of the agent. If no upper bound on the number of tokens is known to the agent, deterministic exploration of all networks is impossible, even if all tokens are fault free. It is also impossible if all tokens are Byzantine, even if their number is known. Our main result is a deterministic exploration algorithm with cost polynomial in the (unknown) size of the network, which works in arbitrary networks, provided that the agent knows some upper bound on the total number of tokens, and that at least one token is fault free.

[1]  Susanne Albers,et al.  Exploring Unknown Environments , 2000, SIAM J. Comput..

[2]  C. Papadimitriou,et al.  Exploring an unknown graph , 1999 .

[3]  Shantanu Das Mobile Agent Rendezvous in a Ring Using Faulty Tokens , 2008, ICDCN.

[4]  Omer Reingold,et al.  Undirected connectivity in log-space , 2008, JACM.

[5]  V. S. Anil Kumar,et al.  Optimal constrained graph exploration , 2001, SODA '01.

[6]  Krzysztof Diks,et al.  Tree exploration with little memory , 2002, SODA.

[7]  Chul E. Kim,et al.  Approximation Algorithms for Some Routing Problems , 1978, SIAM J. Comput..

[8]  Nicola Santoro,et al.  Map construction of unknown graphs by multiple agents , 2007, Theor. Comput. Sci..

[9]  Rudolf Fleischer,et al.  Exploring an Unknown Graph Efficiently , 2005, ESA.

[10]  Michael A. Bender,et al.  The power of team exploration: two robots can learn unlabeled directed graphs , 1994, Proceedings 35th Annual Symposium on Foundations of Computer Science.

[11]  Andrzej Pelc,et al.  Tree exploration with logarithmic memory , 2007, SODA '07.

[12]  Nicola Santoro,et al.  Multiple Mobile Agent Rendezvous in a Ring , 2004, LATIN.

[13]  Masafumi Yamashita,et al.  Erratum: Distributed Anonymous Mobile Robots: Formation of Geometric Patterns , 2006, SIAM J. Comput..

[14]  Adrian Kosowski,et al.  Constructing a Map of an Anonymous Graph: Applications of Universal Sequences , 2010, OPODIS.

[15]  Shantanu Das,et al.  Rendezvous of Mobile Agents When Tokens Fail Anytime , 2008, OPODIS.

[16]  Mona Singh,et al.  Piecemeal graph exploration by a mobile robot (extended abstract) , 1995, COLT '95.

[17]  Michal Koucký Universal traversal sequences with backtracking , 2002, J. Comput. Syst. Sci..

[18]  Andrzej Pelc,et al.  Optimal graph exploration without good maps , 2004, Theor. Comput. Sci..

[19]  Nicola Santoro,et al.  Mobile Agents Rendezvous When Tokens Fail , 2004, SIROCCO.

[20]  Michael A. Bender,et al.  The power of a pebble: exploring and mapping directed graphs , 1998, STOC '98.

[21]  Uri Zwick,et al.  Deterministic rendezvous, treasure hunts and strongly universal exploration sequences , 2007, SODA '07.