Efficient Mechanisms for Exploration of Dangerous Graphs and for Inter-agent Communication

• Main Author: Balamohan, Balasingham
• Language: en
• Published: 2013
• Subjects: Black Hole; Dangerous Graph; Token; Exploration; Map Construction
• Online Access: http://hdl.handle.net/10393/26002

This thesis deals with the problems of exploration and map construction of a dangerous network by mobile agents, and it introduces new general mechanisms for inter-agent communication, which could be applied to other mobile agents' problems. A dangerous network contains a harmful process called Black Hole that destroys all agents entering the node where it resides, without leaving any observable trace. The task for the agents, which are moving asynchronously, is to construct a map of the network with edges incident on the black hole unambiguously identified. Two types of communication mechanisms are considered: whiteboards and tokens. In the whiteboard model every node provides a shared memory on which agents can read and write. When communication occurs through tokens, instead, the agents have some pebbles that can be placed on and picked up from the nodes. Four different costs for comparing the efficiency of the protocols are taken into account: the number of agents required, the number of moves performed, the size of the whiteboard (or the token capacity at a node), and time. The black hole search problem is considered first in ring networks with whiteboards, and optimal exact time and move complexities are established improving all existing results. The same problem is then studied in arbitrary unknown graphs and it is solved in the token model by using a constant number of tokens in total. The protocol improves on existing results and is based on a novel technique for communicating using tokens. Finally, the new method of communicating using tokens described in the context of black hole search is generalized to propose a novel communication mechanism among the agents that could possibly be employed for any distributed algorithm by mobile agents.