Analysis on a mobile agent-based algorithm for network routing and management

Abstract

Abstract-Ant routing is a method for network routing in the agent technology. Although its effectiveness and efficiency have been demonstrated and reported in the literature, its properties have not yet been well studied. This paper presents some preliminary analysis on an ant algorithm in regard to its population growing property and jumping behavior. For synchronous networks, three main results are shown. First, the expected number of agents in a node is shown to be no more than ð1 þ max i fj i jgÞkm, where j i j is the number of neighboring hosts of the ith host, k is the number of agents generated per request, and m is the average number of requests. Second, the expected number of jumps of an agent is shown to be no larger than ð1 þ max i fj i jgÞ. Third, it is shown that for all p ! ð1 þ max i fj i jgÞkm, the probability of the number of agents in a node exceeding p is not greater than R 1 p PðxÞdx, where PðxÞ is a normal distribution function with mean and variance given by Mean = ð1 þ max i fj i jgÞkm, Var. = 2kmð1 þ max i fj i jgÞ +2 ð1þ2 maxifjijgÞ . The first two results are also valid for the case when the network is operated in asynchronous mode. All these results conclude that as long as the value max i fj i jg is known, the practitioner is able to design the algorithm parameters, such as the number of agents being created for each request, k, and the maximum allowable number of jumps of an agent, in order to meet the network constraint.

Analysis and Design of an Agent Searching Algorithm for e-Marketplaces

“…As the explorer agent can only move forward, it finally gets stuck in node A. In order to alleviate this problem, we modify the algorithm presented in [20] as a modified ant algorithm in [17].…”

“…After the search is finished, it will die or randomly jump to either B or D. If it jumps to B and the search at B is finished, it will die or randomly select the next visiting e-marketplace (either A or C). The above jumping mechanism is similar to an ant algorithm in network routing [17,20]. However, this jumping mechanism is one of many methods.…”

“…, 1) T . Since the jumping mechanism of agents is identical to the modified ant routing algorithm that has been presented in [17], the convergence result can thus be expressed by the following theorem.…”

“…If there are too many agents migrating between say A and B, it is for sure that the communication between A and B will be jammed. In [17], we have already proved a bound on the number of propagating agents in the ant routing algorithm. For agent search in e-marketplaces, we can have a similar theorem.…”

“…However, less attention has been paid on the analysis of agent behavior in e-marketplaces. In this paper, we model the jumping (searching) mechanism of agents among the e-marketplace network as a modified ant algorithm [3,17,[19][20][21], which was originally used in network routing. To avoid generating agents using up all resources in the e-marketplace network, we use the existing result [17] of the modified ant algorithm to investigate the behavior of agents in e-marketplaces, as well as to control the population of agents by setting the appropriate agent generation rate.…”