Search

This paper is concerned with the design of a multiprocessor system supporting the parallel execution of functional programs. Parallelism in such systems is automatically derived by the compiler but this parallelism is unlikely to meet the physical constraints of the target machine. In this paper, these problems are identified for the class of Divide-and-Conquer algorithms and a solution which consists of reducing the depth of the computational tree is proposed. A parallel graph reduction machine simulated on a network of transputers, developed for testing the proposed solutions, is described. Experiments have been conducted on some simple Divide-and-Conquer programs and the results are presented. Lastly, some proposals are made for an automatic system that would efficiently execute any problem belonging to the same class taking into account the nature of the problem as well as the physical characteristics of the implementation.

The use of two types of parallel computer hardware for increasing the efficiency of processing in chemical structure data bases is discussed. The distributed array processor can be used for the clustering of 2-D chemical structure data bases by using the Jarvis—Patrick clustering method and for the ranking of output in an experimental system for substructure searching in the 3-D macromolecules in the Protein Data Bank. The Inmos transputer can be used in the construction of PC-based systems for 2-D substructure searching and in the identification of the maximal substructures common to pairs of 3-D molecules.

Self-organisation is one of the most important characteristics of mobile ad hot network (MANET). The performance of protocols proposed for the MANET is usually evaluated by performing extensive simulations. Research has shown that the simulation results are highly dependent on the movement behaviours (mobility models; of mobile nodes. However, most of the mobility models are random-based where some of the important factors related to the simulation environment are not taken into account. In this paper, two realistic mobility models we proposed route mobility model and hotspot mobility model which present the typical environments where road and attractive places exist, respectively. In these models the movements of mobile nodes are either restricted to the pre-defined route or influenced by the hotspot. The results show that the environment factors such as spatial constraints and speed limits may cause significant impact on the network performance.

Bluetooth is a typical short-range radio packet ad-hoc technique that is attractive in the rapid formation of a personal area network and inter-connection of different portable mobile devices. Compared with many high-speed wireless LAN techniques, Bluetooth provides a moderate link speed because each link is only allocated with 1 MHz bandwidth. In order to increase the link utilization, Bluetooth introduces the flexible usage of asymmetric link capacity allocation. It means different speeds are allowed in two directions of a link that flow from the master to the slave and from the slave to the master. Because of this, it is possible for Bluetooth to adaptively adjust the link capacity distribution to increase the link performance with lower latency. In this paper, we use the delay cost function to analyse the capacity assignment for a Bluetooth link. Although it only focuses in solving the capacity assignment of a single link, it is also applicable to large-size multi-link and multi-hop networks. It could be extended to generic data networks with asymmetric links.

Unlike most wireless techniques, Bluetooth uses time division duplex (TDD) to arrange the transmission slots so that it can avoid the collisions. In a Bluetooth piconet, a master node coordinates up to seven active slave nodes, which implies up to fourteen different active virtual links. The schedulers of the devices, especially the one in the master node, are in charge of all the transmission disciplines. Therefore, the policy of the scheduler will affect the networks performance of the piconet, like delay and throughput. Currently the Bluetooth scheduler uses a round robin policy to distribute the slots for different links. An enhancement scheme is presented by introducing the predictive link capacity assignment plus the multi-slot framing. The advantages of the corresponding methods is discussed, which lead to a possible practical design of a new scheduler that exploits a better performance of a Bluetooth piconet.

Medium access control (MAC) is an intermediate layer to manage the policy of link transmissions so that nodes in the same network segment can send packets without constant interference from other nodes. It should give each node the opportunity to transmit packets successfully. Most existing wireless LAN techniques use contention-enable MACs, which are statistical multiplexing algorithms to control multiaccess. Compared to them, regular division-based MACs, such as time division multiplexing (TDM) or frequency division multiplexing (FDM), are seldom used in wireless LANs due to the inefficiency that is shown by the inflexible dynamic bandwidth allocation to nodes. Quality of service (QoS) takes into account this inflexibility in resource reservation in order to grant enough resource to services, even when they are idle. As a result, link connection-oriented networks with mesh topology are considered a good research topic. The corresponding simulation model is required where most current simulation tools do not provide one directly. We present a simple link connection-oriented network simulation design in C without using any existing simulation packages. It can help researchers to design the overall simulation in other related areas, and later it could be extended to a complete simulation tool.

Applications and services for mobile devices can benefit substantially from the use of automatic repeat request (ARQ) link layer protocols for wireless connections. However, a drawback comes with their use when transferring mixed traffic type, which is the introduction of excessive and highly jittered transmission delay, which negatively affect the performance of both real-time and TCP traffic. Here, we propose a selective ARQ protocol which is capable of switching between ARQ and non-ARQ modes based on two policies which we have defined. Our first policy is based on the assumption that for real-time traffic minimisation of delay is more important than reliable transmission. The second policy is an extension of the first and introduces a mechanism, which monitors the length of the buffer queue to decide whether or not packets should be sent with or without ARQ. Results obtained in this research show that real-time traffic can be sent with minimal delay jitter and within certain time boundaries, whilst TCP throughput can be increased significantly.

A self-healing network is a ‘dream’ for every network designer today. Abstractly it is a network that capable of maintaining the availability of network services. This would not be impossible to be turned into a reality if the networks are embedded with intelligence. By having the intelligent capability the network can be self- managed in response to events that taken place within the network itself through the learning mechanisms. This paper presents the User Manager Agent System (UMAS) learning performance in controlling a request for network application session. ANFIS of Matlab v5.3 has been used to simulate the learning process in which it bases on the training data of network state. The work also had revealed some weaknesses of ANFIS in processing a large volume of training data sets and substantial amount of time taken for processing.

An extendible CASE tools environment is currently being researched at the Department of Computer Science, University of Sheffield. This environment is designed primarily for developing parallel system software but is configurable for other applications. It requires an underlying generic data repository to represent information about the system under development in a consistent and complete form. The representation must be independent of the source, and intended use, of its data.

Two chemical substructure searching algorithms, the relaxation algorithm and the set reduction algorithm, are introduced and described. Transputer based serial implementations of both are compared for performance; the relaxation algorithm is shown to be both more effective and more efficient. Strategies are discussed for multi-transputer implementations of the relaxation algorithm. Experimental results show that near-linear speedups are obtained with networks containing up to 21 transputers.

Research activities in the fine chemicals industry involve the generation and analysis of huge amounts of chemical and biological data, and this has led to the development of highly sophisticated, computer-based, chemical information systems to support the research [ 11. Most chemical databases contain two-dimensional structural data, i.e. planar chemical structure diagrams, but there is also increasing interest in the processing of three-dimensional structural data, e.g. from X-ray crystallography and molecular mechanics studies. In this paper, we summarize work in our laboratory over the last 3 years which seeks to determine the utility of parallel hardware for the efficient processing of the structural data that are contained in chemical information systems. Two of the very many different types of parallel computer [2] are discussed, these being the ICL Distributed Array Processor and the INMOS transputer.

It has been argued that security concerns should inform all the stages of the development process of an agent-based system. However, this is not the case since current agent-oriented methodologies do not, usually, consider security concepts within their modelling ontology. In this paper we present extensions to the Tropos ontology to enable it to model security.