Providing reliable agents for electronic commerce

Abstract

Abstract. It is widely agreed that mobile agents in conjunction with WWW technology will provide the technical foundation for future electronic commerce. A prerequisite for the use of mobile agents in a commercial environment is, that agents have to be executed reliable, independent of communication and node failure.In this paper, we first present a recently proposed fault-tolerant protocol to ensure the exactly-once execution of an agent by monitoring the agents execution. With this protocol, agents are performed in so-called stages. Each stage consists of a number of nodes. One of these nodes executes the agent while the other nodes monitor the execution.The main focus of this paper is the construction of stages. In particular, we will investigate how the number of nodes per stage influence the probability of an agent to be blocked due to failures and which nodes should be selected when forming a stage to minimize the overhead caused by the protocol. Also a flexible itinerary concept is proposed that gives agent systems the freedom to do various kinds of optimizations when determining the next node and constructing a stage.

Protecting mobile agents from external replay attacks

“…2, node 2 is a loop node, and the remaining nodes are all regular. However, as proposed in Straßer et al (1998), other nodes types could be defined, for example, to provide more flexibility in the definition of the agent itinerary. These other types would be treated in the same way as regular nodes and, therefore, are not important for the definition of the proposed protection protocol.…”

“…Preventing the agent from being executed more than once has also been referred in the literature as ensuring the exactly-once execution property (Straßer et al, 1998). This property is usually considered when designing fault-tolerant mechanisms for mobile agents.…”

FATOMAS-a fault-tolerant mobile agent system based on the agent-dependent approach

“…We distinguish between spatial-replication-based (SRB) approaches [2,5,9,11] and temporal-replication-based (TRB) [6,10] approaches; our paper advocates a SRB approach.…”

“…To ensure the exactlyonce execution property, SRB schemes need to solve an agreement problem. Current SRB schemes assume reliable failure detection [5], are based on complex models [2,9,11], or block even on a single place failure [9,11] (see [8] for an in-depth discussion). Our approach, which is based on an easily understandable model, does not assume reliable failure detection and prevents blocking.…”

System mechanisms for partial rollback of mobile agent execution

“…The set of actions an agent has to perform on a single node is called a step and is implemented as a single method of the agent object. Which step the agent has to perform on which node and the order in which the steps have to be performed is described by an itinerary, which may be adapted during the execution of the agent [15]. If an agent migrates to another node, the agent object with code and all private data belonging to the object is captured and transferred to the next node.…”