Browsing by Author "K., Pradheep Kumar"

Now showing 1 - 10 of 10

Application of non-uniform laxity to EDF for aperiodic tasks to improve task utilisation on multicore platforms
(ARXIV, 2009-06) K., Pradheep Kumar
This paper proposes a new scheduler applying the concept of non-uniform laxity to Earliest deadline first (EDF) approach for aperiodic tasks. This scheduler improves task utilisation (Execution time / deadline) and also increases the number of tasks that are being scheduled. Laxity is a measure of the spare time permitted for the task before it misses its deadline, and is computed using the expression (deadline - (current time + execution time)). Weight decides the priority of the task and is defined by the expression (quantum slice time / allocated time)*total core time for the task. Quantum slice time is the time actually used, allocated time is the time allocated by the scheduler, and total core time is the time actually reserved by the core for execution of one quantum of the task. Non-uniform laxity enables scheduling of tasks that have higher priority before the normal execution of other tasks and is computed by multiplying the weight of the task with its laxity. The algorithm presented in the paper has been simulated on Cheddar, a real time scheduling tool and also on SESC, an architectural simulator for multicore platforms, for upto 5000 random task sets, and upto 5000 cores. This scheduler improves task utilisation by 35% and the number of tasks being scheduled by 36%, compared to conventional EDF.
Catur Approach to Assess the Quality of Big Data Using Decision Tree and Multidimensional Model
(AENSI Publisher, 2015) K., Pradheep Kumar
This paper is intended to design and develop multidimensional and decision tree based frameworks, for assessing the quality of a big data. Since the datasets represented in a big data environment is both complex and multidimensional, the quality of big data can be better viewed through multiple dimensions. Most enterprises face number of challenges in managing the quality of the big data during their initial setup or migration from traditional database or after building the big data. This paper uses multidimensional model proposed for Knowledge Management System for designing critical quality dimensions for big data. Based on the extensive literature review, this work proposes a classification of big data quality into many quality factors such as accessibility, consistency, integrity, usability, relevance, completeness, compatibility, conformity and accuracy. Since there are very few appropriate data stewards or frameworks available for confirmation of quality dimensions, this paper aims to develop some hybrid approaches using multi-dimensional model and decision tree based methods for automatic quality checks. Using decision tree, multiple if-then rules can be formed to decide on the quality of data based on the specific constraints developed for big data. The paper also aims to provide the quality framework and measures which can serve as a data quality firewall just like an internet firewall to proactively find the quality issues and apply the rules based on the decision tree algorithms to prevent bad or inconsistent or invalid data or access entering in to the big data environment.
Fuzzy Based Augmented Reality for 3D Image Modelling
(TEST Engineering & Management, 2020) K., Pradheep Kumar
In today’s world Augmented Reality and Virtual Reality is of prime importance. To create a scenario using Augmented Reality it is important to model objects in 3D space. Once the modelling is complete the Augmented Reality Map could be used in several applications like medicine where 3D bio printing should be done. It could also be used in education and teaching to illustrate complex working mechanisms. Here a fuzzy based algorithm has been proposed to create 3D models of objects for Augmented Reality maps. The Fuzzy rule method reduces RMSE, compared to AR Marker, Fingertips and Checkerboard by 35%, 45% and 21% respectively. The Fuzzy rule method also improves accuracy of resolution of images, compared to AR Marker, Fingertips and Checkerboard by 48%, 11% and 11% respectively.
Fuzzy based modeling for an effective IT security policy management
(IEEE, 2016) K., Pradheep Kumar
This paper presents a fuzzy based modeling approach for determining the effectiveness of IT security policies of an organization. Normally organizations have different IT security policies, but modeling and maintaining them is a tedious process. In order to ensure appropriate IT support policies, there are many factors involved. The most common are the periodic data backups and prevention against the virus attack. Therefore, it is essential to ensure that the IT security policies are assessed for all the possible vulnerabilities and hidden threats. Thus fuzzy based modeling has been proposed to identify the effectiveness of the IT security policies for any organization. Mamdani's Fuzzy inference engine has been applied to construct the fuzzy rules based on the different primary and secondary factors. Finally the results from the analysis of the factors are processed by the fuzzy inference engine, to decide the rule strengths and its effectiveness of the overall IT security policies.
Fuzzy-Based Querying Approach for Multidimensional Big Data Quality Assessment
(2017) K., Pradheep Kumar
This paper is intended to design a fuzzy based approach to assess standards and quality of big data. It also serves as a platform to organizations that intend to migrate their existing database environment to big data environment. Data is assessed using a multidimensional approach based on quality factors like accuracy, completeness, reliability, usability, etc. These factors are analysed by constructing decision trees to identify the quality aspects which need to be improved. In this work fuzzy queries have been designed. The queries are grouped as sets namely Excellent, Optimal, Fair and Hybrid. Based on the fuzzy data sets formed and the query compatibility index, a query set is chosen. A data set that has a very high degree of membership is assigned a fair query set. A data set with a medium degree of membership is assigned a optimal query set. A data set that has a lesser degree of membership is assigned a Excellent query set. A data set which needs a combination of queries of all the above is assigned a hybrid query set. The fuzzy query based approach reduces the query compatibility index by 36%, compared to a normal query set approach.
Modified Backward Chaining Algorithm Using Artificial Intelligence Planning IoT Applications
(IGI Global, 2019) K., Pradheep Kumar
In this chapter, an automated planning algorithm has been proposed for IoT-based applications. A plan is a sequence of activities that leads to a goal or sub-goals. The sequence of sub-goals leads to a particular goal. The plans can be formulated using forward chaining where actions lead to goals or by backward chaining where goals lead to actions. Another method of planning is called partial order planning where all actions and sub-goals are not illustrated in the plan and left incomplete. When many IoT devices are interconnected, based on the tasks and activities involved resource allocation has to be optimized. An optimal plan is one where the total plan length is minimum, and all actions consume similar quantum of resources to achieve a goal. The scheduling cost incurred by way of resource allocation would be minimum. Compared to the existing algorithms L2-Plan (Learn to Plan) and API, the algorithm developed in this work improves optimality of resources by 14% and 36%, respectively
Multicore real time scheduling using fuzzified priority and non-uniform laxity
(IEEE, 2010) K., Pradheep Kumar
Scheduling of tasks takes into account the priority of the tasks and the available laxity among other factors. Laxity is a measure of slack time. Generally, the available slack time is cumulated and uniformly distributed among all tasks, irrespective of priority. However, a better approach would be to distribute the slack time to tasks in proportion to their priority, giving rise to non-uniform laxity distribution. Also, these parameters may not always be specified precisely. In such situations, these parameters can be specified as approximate linguistic variables and handled through the fuzzy inference engine. Fuzzification is used as a tool to obtain the execution eligibility from the input linguistic variables. The execution eligibility is the order in which tasks are dispatched to the execution queue. Based on the execution eligibility, the tasks are scheduled. The remaining time, if available after scheduling the task, is cumulated as slack time. A task is always scheduled based on minimum execution time allotted. If a task misses its deadline due to inadequate resources, the difference between scheduled time i. e., (Arrival time + minimum execution time) and deadline is computed. If this amount of time is available in the cumulated slack, a relative deadline is set by adding the deadline to the slack needed, and the task is scheduled. If slack is not available in the cumulated slack, the task misses its deadline. The objective of this approach is to ensure that a task with higher execution eligibility does not miss its deadline, when slack time is available. The algorithm presented in this paper has been simulated on Cheddar, a real time scheduling tool, and also on SESC, an architectural simulator for multicore platforms.
Pfairness Applied to EDF to Reduce Migration Overheads and Improve Task Schedulability in Multicore Platforms
(IEEE, 2009) K., Pradheep Kumar
This paper proposes a scheduler combining the concepts of EDF and pfairness using the worst fit heuristic function. In scheduling algorithms without pfairness, priority is not monitored closely in case of preemptions. An algorithm combining EDF and pfairness proposed in this paper overcomes this drawback. Here resources are granted in accordance to the task weight. Individually using either EDF or pfairness utilizes the resources to a greater extent, whereas a combination of both achieves better reduction in migration overheads. The algorithm has been simulated on Cheddar, a real time scheduling tool, and also on SESC, an architectural simulator on multicore platforms. The algorithm presented in this paper has been tested for 5000 random task sets. The results show that it reduces the migration overhead by 33% for partitioned task sets and by 38 % for hybrid task sets, and improves task schedulability by 37%, compared to conventional EDF.
Scheduling Algorithm for Asymmetric Multicore Platforms using Non-uniform Laxity based Clustering and Duplication
(European Journal of Scientific Research, 2012-10) K., Pradheep Kumar
In this paper, a new scheduling algorithm using task clustering and non-uniform laxity is proposed. This algorithm forms task clusters after analyzing the task dependencies. For each task cluster comprising of a parent and dependent tasks, a cluster parameter is computed. Based on this parameter, a rank is assigned to each cluster. The clusters are scheduled in the order of their rank, so as to minimize the communication cost and reduce the utilization of slack resources. Clusters are scheduled as far as possible on the same core. If a cluster cannot be scheduled completely on the same core, the parent task is either migrated or duplicated depending on the associated costs involved. The algorithm has been tested using 15 different programs as tasks.
Shortest Distance Lattice Cryptographic Algorithm for Data Points Using Quantum Processors
(Springer, 2021-08) K., Pradheep Kumar
In this paper, a Lattice cryptographic algorithm has been proposed by computing the shortest distance between points in clusters. It classifies points which are dependent and independent. Subgrouping of these points as clusters are formed comprising of parent and multiple child points. Proximity based on similarity is computed as a measure of distance between points. The distance between parent points and child points are also computed and arranged in increasing order. Another subgrouping is made with the points based on this arrangement. This module is called as the cryptographic unit. Later data points are grouped separately and distances are computed for various combinations and arranged in increased. The points are encrypted and later decrypted based on simple matrix operations. The average processing time of SDL has been 55 s and it gives a reduction of 13% as compared to ECC and 23% as compared to RSA.