Dr. K. M. Azharul Hasan Professor Department of Computer Science & Engineering Khulna University of Engineering & Technology (KUET) Khulna –9203, Bangladesh. Phone: +880-41-769471 Ext. 355 (Personal) 350 (Office) Mobile: +880-1714087273 Fax: +880-41-774403 Email: azhasan@gmail.com , azhasan@cse.kuet.ac.bd Web: www.kuet.ac.bd/cse/azhasan
Abstract: Scheduling for data exchange in Vehicular Ad Hoc Networks (VANETs) is being considered as a vital issue by the researchers. Road Side Units (RSUs) being the standalone buffer point is to compensate the less connectivity problem due to mobility of vehicles. Here we formulate an RSU-based VANETs model. Due to vehicle mobility and RSU range constraint scheduling is an important issue to meet a vehicle’s issued demand in hard real-time. We investigate existing scheduling algorithm serve a data item only based on vehicles’ submitted requests’ characteristics and avoiding the relationship between a vehicle and its submitted requests which leads a vehicle level starvation problem. In this paper, we address this starvation problem and propose a solution to minimize the problem. Simulation result also supports our proposed approach and offers expected result.
Abstract: Cohesion and coupling are internal software attributes, which tell how tightly the components of software module are bound together in design or implementation. Cohesion and coupling are the desirable goal in software construction, leading to better values for external attributes such as maintainability, reusability, and reliability. Object-oriented (OO) system development is gaining wide attention both in research environment and in industry. A severe problem is encountered, however, is the quickly increasing complexity of such systems and the lack of adequate criteria and guidelines for good designs. To cope with this problem it is imperative to better understand the properties and characteristics of object-oriented systems. Cohesion describes the binding of elements within one method and within one object class, respectively. Coupling describes the interdependency between methods and between object classes, respectively. Considering all the features of object-orientation the coupling in object-oriented system has been categorized. A model has been presented for measuring direct coupling of OO system, which is then used to identify components with the objective of encapsulating higher coupling objects into the components. Some ways has been suggested for which an OO approach can be developed with high cohesion and low coupling. Relationships between two well-known and widely accepted measures of cohesion have been shown. The effects of Public members, Private members, Constructors, Destructors and Inheritance in measuring cohesion have been analyzed through principal component analysis. Some statistical analysis has been shown to validate the proposed measures.
Abstract: New application areas for databases like data warehousing and OLAP (Online Analytical Processing) deploy the multidimensional data model in order to describe the application domain. Consequently, OLAP systems are represented by a multidimensional database schema to adequately reflect the application semantics. In this thesis a new implementation scheme of relational table for multidimensional databases called HORT is presented based on multidimensional extendible arrays. The data allocation is a key performance factor for MOLAP systems. This holds especially for data warehousing environments where huge amounts of data have to be dealt with. In this study a new data allocation scheme is used based on the notion of subarray. Using this allocation it is shown that the storage and retrieval cost has significant improvement over the conventional implementation of relational table. A new parallel implementation scheme of relational tables for multidimensional databases named P-HORT is presented. The P-HORT scheme uses a new partitioning scheme based on the notion of subarray. The superiority of this partitioning comparing to other existing schemes is shown. In P-HORT partitioning scheme the candidate and non candidate records for a query are separated. The effectiveness of extendible array to be implemented in parallel environment is shown by pointing out some interesting properties of extendible arrays. Static and dynamic load balancing for the parallel implementation is developed. The dynamic load balancing shows considerable performance for P-HORT implementation. Cost models for both static and dynamic load balancing is developed. Sufficient experimental results are presented and the cost models are validated. Efficient MOLAP implementation to manage array extension by keeping track of the extension subarrays is proposed. New data structure is developed to apply in OLAP applications. Cost analysis is presented showing the significant savings both in retrieval and extension using extendible array.
