Mohammad Mursalin Akon

A model for designing and implementing parallel applications using extensible architectural skeletons

With the advent of hardware technologies, high-performance parallel computers and commodity clusters are becoming affordable. However, complexity of parallel application development remains one of the major obstacles towards the mainstream adoption of parallel computing. As one of the solution techniques, researchers are actively investigating the pattern-based approaches to parallel programming. As re-usable components, patterns are intended to ease the design and development phases of a parallel applications. While using patterns, a developer supplies the application specific code-components whereas the underlying environment generates most of the code for parallelization. PAS (Parallel Architectural Skeleton) is one such pattern-based parallel programming model and tool, which defines the architectural aspects of parallel computational patterns. Like many other pattern-based models and tools, the PAS model was hampered by its lack of extensibility, i.e., lacking of support for the systematic addition of new skeletons to an existing skeleton repository. Lack of extensibility significantly reduces the flexibility and hence the usability of a particular approach. SuperPAS is an extension of PAS that defines a model for systematically designing and implementing PAS skeletons by a skeleton designer. The newly implemented skeletons can subsequently be used by an application developer. SuperPAS model is realized through a Skeleton Description Language (SDL), which assists both a skeleton designer and an application developer. The paper discusses the SuperPAS model through examples that use the SDL. The paper also discusses some of the recent usability and performance studies, which demonstrate that SuperPAS is a practical and usable parallel programming model and tool.

Modeling Epidemic Data Diffusion for Wireless Mobile Networks

We propose an analytical model for data diffusion time/delay in a wireless mobile network using a novel peer-to-peer spatial-demand based information dissemination technique. The demand-based technique for data dissemination is beneficial for mobile network since this fully distributed and scalable network system utilizes only local urge for data and provides faster delivery of information. However, due to mobility and chaotic wireless network, it is difficult to predict the object diffusion time/delay among all the interested nodes in a mobile network. Therefore, the development of an analytical model to anticipate the expected time of data distribution among the nodes in a mobile system is an important area of research. In response to this problem, we first find the probabilities of transmitting object from one node to multiple nodes using the epidemic model of disease spreading. Utilizing these transition probabilities, we construct an analytical model based on Markov chain to calculate the expected delay of information diffusion. In addition, we adopt the mobility and scheduling impact on data transition probabilities in our analytical model. Extensive event-based simulations demonstrate that our analytical model provide near perfect estimation of data diffusion time/delay in wireless mobile networks.

An inexpensive unstructured platform for wireless mobile peer-to-peer networks

In this paper, we propose an unstructured platform, namely InexpensivePeer-to-PeerSubsystem (IPPS), for wireless mobile peer-to-peer networks. The platform addresses the constraints of expensive bandwidth of wireless medium, and limited memory and computing power of mobile devices. It uses a computationally-, memory requirement- and communication- wise inexpensive gossip protocol as the main maintenance operation, and exploits location information of the wireless nodes to minimize the number of link-level messages for communication between peers. As a result, the platform is not only lightweight by itself, but also provides a low cost framework for different peer-to-peer applications. In addition, further enhancements are introduced to enrich the platform with robustness and tolerance to failures without incurring any additional computational and memory complexity, and communication between peers. In specific, we propose schemes for a peer (1) to chose a partner for a gossip iteration, (2) to maintain the neighbors, and (3) to leave the peer-to-peer network. Simulation results are given to demonstrate the performance of the platform.

Routing in telecommunication network with controlled ant population

Swarm intelligence based algorithms for routing in telecommunication networks have gained considerable amount of attention in recent years. Swarm intelligence, demonstrated by natural biological swarms, exhibits numerous powerful features that are desirable in communication systems. Mobile agents having swarm intelligence similar to real life ants have been employed to achieve a general, flexible and modular solution for routing in communication networks. We propose a novel scheme to control a population of ant-like mobile agents that can ultimately result in better load balancing and better performance within the network.

A bandwidth and effective hit optimal cache scheme for wireless data access networks with client injected updates

In this paper, we propose an optimal cache replacement policy for data access applications in wireless networks where data updates are injected from all the clients. The goal of the policy is to increase effective hits in the client caches and in turn, make efficient use of the network bandwidth in wireless environment. To serve the applications with the most updated data, we also propose two enhanced cache access policies making copies of data objects strongly consistent. We analytically prove that a cache system, with a combination of our cache access and replacement policy, guarantees the optimal number of effective cache hits and optimal cost (in terms of network bandwidth) per data object access. Results from both analysis and extensive simulations demonstrate that the proposed policies outperform the popular Least Frequently Used (LFU) scheme in terms of both effective hits and bandwidth consumption. Our flexible system model makes the proposed policies equally applicable to applications for the existing 3G, as well as upcoming LTE, LTE Advanced and WiMAX wireless data access networks.

SPID: A Novel P2P-Based Information Diffusion Scheme for Mobile Networks

In this paper, we propose a novel peer-to-peer based information diffusion scheme for mobile ad hoc networks. Due to mobility and bandwidth constraints, it is not always possible to share a complete content within a single transmission from one mobile device to another. To address this problem, we introduce segmentation of contents into smaller pieces for efficient information exchange. In mobile ad hoc networks, since it is impossible for a node to know the content of every other node in the network, we propose a Spatial-Popularity based Information Diffusion (SPID) scheme that determines urgency of dissemination of content according to the necessity in a neighborhood. Substantially, the segmentation policy and popularity based information dissemination reduce content acquiring time for each peer. Extensive simulation studies demonstrate that the proposed scheme is efficient in data distribution in mobile ad hoc networks, and prove superiority to other existing methods.

OUR: Optimal Update-based Replacement policy for cache in wireless data access networks with optimal effective hits and bandwidth requirements

In mobile wireless data access networks, remote data access is expensive in terms of bandwidth consumption. An efficient caching scheme can reduce the amount of data transmission, hence, bandwidth consumption. However, an update event makes the associated cached data objects obsolete and useless for many applications. Data access frequency and update play a crucial role in deciding which data objects should be cached. Seemingly, frequently accessed but infrequently updated objects should have higher preference while preserving in the cache. Other objects should have lower preference or be evicted, or should not be cached at all, to accommodate higher-preference objects. In this paper, we proposed Optimal Update-based Replacement, a replacement or eviction scheme, for cache management in wireless data networks. To facilitate the replacement scheme, we also presented two enhanced cache access schemes, named Update-based Poll-Each-Read and Update-based Call-Back. The proposed cache management schemes were supported with strong theoretical analysis. Both analysis and extensive simulation results were given to demonstrate that the proposed schemes guarantee optimal amount of data transmission by increasing the number of effective hits and outperform the popular Least Frequently Used scheme in terms of both effective hits and communication cost. Copyright

SuperPAS: a parallel architectural skeleton model supporting extensibility and skeleton composition

Application of pattern-based approaches to parallel programming is an active area of research today. The main objective of pattern-based approaches to parallel programming is to facilitate the reuse of frequently occurring structures for parallelism whereby a user supplies mostly the application specific code-components and the programming environment generates most of the code for parallelization. Parallel Architectural Skeleton (PAS) is such a pattern-based parallel programming model and environment. The PAS model provides a generic way of describing the architectural/structural aspects of patterns in message-passing parallel computing. Application development using PAS ishierarchical, similar to conventional parallel programming using MPI, however with the added benefit of reusability and high level patterns. Like most other pattern-based parallel programming models, the benefits of PAS were offset by some of its drawbacks such as difficulty in: (1) extending PAS and (2) skeleton composition. SuperPAS is an extension of PAS that addresses these issues. SuperPAS provides a skeleton description language for the generic PAS. Using SuperPAS, a skeleton developer can extend PAS by adding new skeletons to the repository (i.e., extensibility). SuperPAS also makes the PAS system more flexible by defining composition of skeletons. In this paper, we describe SuperPAS and elaborate its use through examples.

Hit Optimal Cache for Wireless Data Access

One of the working requirements for many data access applications is the availability of the most updated information. In wireless communications service networks, remote data access consumes expensive wireless spectrum. An efficient cache can reduce the data access cost, by cutting down the amount of data transferred over the wireless channels. However, deploying an efficient cache in wireless environment is challenging, where caches are distributed all over the network and an update event invalidates all the cached copies of the updated data. In this paper, we formalize the concept of caching frequently accessed but infrequently updated data objects, and propose a cache replacement policy accordingly. To facilitate the replacement policy, two enhanced cache access policies are also proposed. The proposed caching policies are supported with strong theoretical analysis. We demonstrate that the policies guarantee an optimal number of cache hits in a caching system. Results from both analysis and our extensive simulations demonstrate that the proposed policies outperform the popular Least Frequently Used (LFU) scheme in terms of effective hits.

Low Cost Peer-to-Peer Collaborative Caching for Clusters

Retrieving data from the memory of a remote computer is significantly faster than retrieving data from a remote storage device. Therefore, sharing caches among several computers can reduce the number of storage device access. In this paper, we propose a new collaboration scheme among clients in a computer cluster to share their caches with each other. The collaboration is designed based on peer-to-peer computing model. The scheme not only provides an easy to develop software architecture but also yields a near optimal performance with rea- sonably low overheads. Simulations are conducted to demonstrate the performance of the proposed scheme.