RESEARCHED AND ABSTRACTS
Researcher: Hitham Alhussain, Nordin Zakaria, Lukman Abd Rahim and Izzatdin Abd Aziz
Title: Towards Real Time Scheduling in the Cloud
Dynamic provisioning of virtual machines enables a cloud platform to adapt to workload demands. In this paper, we present an architecture and algorithms that exploit this capability for the support of soft real-time tasks scheduling in cloud computing environments. Three soft real-time task scheduling algorithms are modified and integrated into the proposed architecture: Earliest Deadline First (EDF), Earliest Deadline until Zero-Laxity (EDZL), and Unfair Semi-Greedy (USG). A deadline look-a-head module is incorporated into each of the proposed algorithms, that fires deadline exceptions priory in an attempt to avoid missing deadlines and maintain the system criticality. The obtained results are presented in-terms of average deadline exceptions - the extra resources consumed by each algorithm to handle the deadline exceptions - and the average response time. Specifically, the obtained results not only suggest that soft real-time scheduling of periodic real-time tasks is attainable in cloud computing, but it might also scale-up to handle near hard real-time task scheduling.
Researcher: Guilherme Dinis Junior and Nordin Zakaria
Title: Dynamic scheduling on an open-source based volunteer computing infrastructure
Distributed computing can help realize high computational power, through the gathering of dispersed resources. Nowadays, there are different types of computing infrastructures, like grids and volunteer communities, which have distinct needs. The preservation of energy or other resources is chief among them. To address these problems, novel scheduling algorithms can be used; recent work has placed great emphasis on energy-aware scheduling, but literature shows that many others focus on different fields, such as data-aware scheduling. In order to realize these algorithms as solutions, distributed middleware systems must be capable of supporting these scheduling policies; alas, doing so requires considerable development effort for each solution. Towards this end, we have incorporated dynamic scheduling into an existing open-source volunteer computing system named BOINC. Our aim was to start a path into what could be a collection of middleware tools with standardized integration to sensor systems to support distinct types of scheduling policies, without the need to recompile the server system.
Researcher: Soodeh Peyvandi and Rohiza Ahmad
Title: Reliability and Availability of Volunteers in Desktop Grid Computing Environment
Scientific and engineering projects such as SETI@home and Distributed.net are consuming resources of distributed internet-connected idle computers, known as volunteers, for executing or computing their jobs. This type of large distributed system is referred to as desktop grid(DG) computing system or volunteer computing (VC) system. In desktop grid computing environment, a huge scientific project will be broken down into small independent units or tasks and passed to some available idle volunteers for parallel execution. However, these volunteers have one major issue, which is volatility. In other words, their availability throughout job execution cannot be guaranteed and is up to the jurisdiction of the owners. Hence, resource allocation and job scheduling are two main components which are very critical for a desktop grid system. Both have important roles in ensuring efficient, effective and successful execution of jobs. On top of that, other factors need to be considered as well - heterogeneous nature of the resources. In this research, we seek to develop an availability scoring model for volunteer computing. The model is to be deployed during resource allocation and job scheduling in the desktop grid environment. The availability scoring model is formulated based on the results obtained from statistical analysis of historical trace data set. Based on the availability score obtained, the heterogeneous volunteers are then grouped into four homogeneous groups, each with a different reliability level.
Researcher: Ali Usman Abdullahi, Rohiza Ahmad and Nordin Zakaria
Title: Framework for OLAP Based Map Reduce High Level Query Languages.
The value of Big Data largely relies on its analytical outcomes; and MapReduce has so far been the most efficient tool for performing analysis on the data. However, the low level nature of MapReduce programming necessitates the development of High-level abstractions, i.e., High Level Query Languages (HLQL), such as Hive, Pig, JAQL and others. These languages can be categorized as either dataflow based or OLAP-based. For OLAP-based HLQL, in particular Hive, at the moment, the speed of retrieval of big data for the analysis is still requiring improvement. Hence, indexing is one of the techniques used for this purpose. Yet, the indexing approach still has its loopholes since it is performed manually and externally using the approach of index inclusion and two-way data scanning. It requires huge computational time and space and hence not scalable for future potential scale of big data. Thus, an adaptive indexing framework is proposed for improving both the computational time and memory usage of the indexing process. The technique shall parse user queries to determine the necessity for indexing and use internal indexing with one-way data scanning approach for the indexing strategy.
Researcher: Muhammad Usama and Nordin Zakaria
Title: Chaos-based Secure Data Compression for Hadoop
With the increasing need to store and process confidential data in the cloud, compression and encryption capability will be increasingly critical in commonly deployed distributed computing platforms such as Hadoop. While multitudes of data compression and encryption tool are already available in such platforms, the tools are typically applied in a sequence compression followed by encryption, or encryption followed by compression. This paper focusses on Hadoop and proposes a data storage method for it that effectively and efficiently couple data compression and encryption. This method incorporates an improved version of GLS-Coding, Range GLS-Coding. The improved version solves the implementation issue of infinite precision real numbers requirement in the original algorithm by removing the fractional part generated by long products of real numbers. It incorporates a secret key by shifting the direction and the performing cyclic shift operation in a piecewise linear chaotic map (PWLCM) without compromising its compression capabilities. Further, encryption quality is enhanced by the introduction of masking pseudorandom keystream. The proposed algorithm was shown to fit in well within the Hadoop framework, providing competitive encryption security and compression at the same time to the data stored.
Researcher: Ahmad Abba Haruna, Low Tan Jung and Nordin Zakaria
Title: Cooling-Efficient Job Scheduling in a Heterogeneous Grid Environment
Electricity consumption typically forms the biggest portion of a Data Center’s operational cost, with the biggest consumers, in roughly equal proportion, being the servers and the cooling units. In an effort to reduce electricity consumption,we propose a grid scheduling algorithm that takes advantage of a Gas-District Cooling Data Center model to reduce the cooling energy consumption. Experimental analysis shows that the proposed method was able to reduce cooling electricity consumption by 20%. Further, by increasing the maximum allowable temperature by 1 degree, the proposed method was able to save an additional 3%.
Researcher: Saddaf Rubab, Mohd Fadzil Hassan, Ahmad Kamil Mahmood, Syed Nasir Mehmood Shah
Title: Performance Management Using Autonomous Control-Based Distributed Coordination Approach in a Volunteer Grid Computing Environment
In volunteer grid environment, it is difficult to fulfill the requirements of all jobs due to increasing demands of resources. A resource requester submits a job, require resources for the job to be completed within deadline and budget if specified any. Whereas resource provider makes use of available resources and wants to utilize resources to maximum. Therefore, satisfying the requirements of both i.e., jobs and resources makes it difficult to manage the performance of a volunteer grid. In performance management, the main objectives include maintaining service level agreements, maximization of resource utilization, meeting job deadline/budget and minimizing the job transfer. In this paper, only the maximization of resource utilization and meeting job deadlines will be addressed for managing the performance of a volunteer grid computing environment. An autonomous approach is introduced that provides dynamic resource allocation for submitted jobs in a volunteer grid environment depending on the availability and demand of resources. Grid resource brokers are considered third party organizations that work as intermediaries between volunteer resource provider and requester. Proposed autonomous approach is developed by utilizing distributed coordination approach for interactive assignment of volunteer resources. The proposed approach is applying distributed coordination approach and giving priority to maximization of volunteer resource usage while completing jobs within deadline.
Researcher: Shiladitya Bhattacharjee and Lukman Abd Rahim
Title: A Hybrid Technique in Achieving Confidentiality, Integrity and Robustness for Big Data Transmission
Data loss, robustness, integrity and confidentiality issues in big data transmission occurs due to extra overhead, channel congestion, transmission errors, and diverse security attacks. Plentiful conducted researches to address these issues show that the applications of various security techniques to enhance data confidentiality increase data overhead. Conversely, an efficient data compression liquefies data confidentiality. Henceforth, this research has been conducted to address these issues collectively with no negotiation of any security aspect. The confidentiality issues are addressed and a backup system for accidental data loss is offered with the combination of SDES encryption and an advanced pattern string generation technique, where a distinct table is used to generate pattern string. Consequently, the robustness against various transmission errors is improved by incorporating a novel error control technique, which encompasses dual round of XOR operations for generating and incorporating error control bits. A modern fixed length based compression is used to address data portability and integrity issues uniquely. Confidentiality and integrity are further intensified by integrating an advance LSB based audio steganography which uses a distinctive sample selection for hiding bits. Experiments are conducted to validate the proposed method using standard Calgary Corpuses and diverse texts and WAV audio files of various sizes. Result shows that the proposed technique is efficient to offer confidentiality as it offers higher Signal to Noise Ratio, lower Amplitude Difference and Frequency Difference than existing. The achieved results displayed its superiority to protect different security attacks as it offers higher Avalanche Effects and Entropy Values than other corresponding. It is also efficient to offer higher integrity and robustness as it offers higher Signal to Noise Ratio, lower percentages of Information Loss and Uncorrectable Error Rate than others. Furthermore, it is capable to offer higher portability than others as it produces lower Bits per Code, higher Compression Percentage, lower Compression Ratio and higher percentages of Space Saving. It is time competent than others too as it offers higher Throughput than exiting. Hence, the proposed hybrid technique can contribute significantly where high security is needed to transfer big data, required, for instance, banking, financial sector and research organizations.