Study On Fault Tolerance Of Delay Tolerant Networks Based On Bio-Network

Delay Tolerant Network (DTN) is a class of networks designed for satellite communication networks, wireless Ad-Hoc networks, sensor networks, transport networks, and networked control systems etc. It is originally produced by Delay Tolerance Network Study Group (DTNRG) of Internet Research Task Force (IRTF). DTN experiences frequent and long-duration partitions due to sparse distribution of nodes. It presents additional difficulties to network congestion control and flow control. And current networking technologies rely on a set of fundamental assumptions that are not true in DTN environments.Fault tolerance is an important method in network systems. Research of fault tolerance covers a wide spectrum of applications, including system architecture, design techniques, coding theory, testing, validation, software reliability, and real-time processing, etc. However, some problems in DTNs, such as intermittent connectivity due to power scheduling, node failure, and packet losses from unpredictable external factors, are frequently encountered. The pure fault tolerant technologies are limited, as such more intelligent and practical methods are required. In this thesis, based on intelligent mechanisms of bio-network, some reliable problems in DTNs and supervisory systems are studied.Firstly, we present a fault tolerant scheme to solve path failure problems. In DTNs, path failure occurs frequently, so message transfer is not reliable. Some well-known assumptions of traditional networks are no longer true in DTNs. We describe the path failure process in DTNs when the path appears completely normal, completely failed and partially failed. Traditional approaches by using precisely known network dynamics have not accounted for message losses. We adopt a fault tolerant scheme to generate redundancy by using erasure coding and full replication. This can greatly decrease the path failure rate. A traffic DTN model is analyzed and the results reveal the superiority of our scheme in comparison to other present schemes.Secondly, from theories and studies of DTNs, we design a DTN oriented intelligent bio-network architecture. The architecture includes distributed terminal layer, bio-network layer, extended layer, mechanism layer, and application interface layer. Network services required are emergenced from congregated biological agents. The service emergence can provide a novel solution to design intelligent service and application in DTNs.Again, a new custody transfer scheme based on the bio-network is proposed. Custody transfer is a proposed DTN reliability mechanism to enhance end-to-end reliability. This mechanism utilizes hop-by-hop transfer of delivery responsibility. This thesis presents the custody transfer mechanism facing problems of duplicate message fragments and limited storage resources, which can prevent further traffic flowing even when some outgoing connections are available. A custody transfer model is built on the bio-network platform to examine its performance of this new scheme that it can increase the end-to-end reliability. An application of garment-embedded oriented wireless sensor network is analyzed. Results reveal the superiority of the Bio-Network scheme in comparison to the present scheme.Moreover, we present a channel fault tolerant structure based on the bio-network and design the features of biological agents. Biological agent delegates channel service. Then, biological agent is extended to support distributed awareness. Considering the connect affinity, multiple affinity-based service quality model, and trust affinity, the concrete methods for service awareness are supported. The simulation results show that the approach is of high performance and extendable capability.Next, concerned with the problems of low automation level and high operation cost for supervisory systems, a layered bio-network is studied and the biological agent as its functional unit is analyzed. In the bottom layer, the communication infrastructure is built to hide the difference among underlying heterogeneous systems. In the bio-system layer, each biological agent delegates system service to gain the functions. Then, in the application layer, various services are encapsulated upon biological agents. The application interface makes the integration with other systems more convenient. Based on the bio-network, a novel framework of supervisory system for typical waterworks is proposed. The simulation results prove that the bio-network based system performs better in terms of response time over the tranditional system.Finally, a summary of this thesis is made, and the further developments are narrated.