Search

Fog computing has emerged as the most popular technology for processing delay-sensitive tasks in the Internet of Things platform. However, offloading tasks to suitable fog nodes (FNs) for their processing is a very challenging problem which is known to be NP-hard. In this paper, we propose a metaheuristic-based task offloading scheme that jointly optimizes the offloading delay and resource utilization. We first formulate the task offloading as a multi-objective optimization problem and then propose an algorithm based on the Discrete Jaya Algorithm (DJA) to solve it. The DJA is presented with efficient candidate representation and mutation operation. We also propose a task migration algorithm to transfer the partially completed task to another FN before failure. Through extensive simulations, we demonstrate the efficacy of the proposed work by comparing the state-of-the-art algorithms and validating it through the Friedman test. [ABSTRACT FROM AUTHOR]

The stability of a feature selection (FS) algorithm is one of the most crucial issues when working with a machine learning model. Until now, various stability measures based on a subset of features have been proposed. However, they lack consideration for feature ranking which is equally important to judge the robustness of algorithms. This paper proposes a novel frequency-based stability measure called rank stability (RSt) that evaluates FS algorithms on both criteria, i.e., subsets of features and feature rankings. The proposed measure evaluates the variation of feature rankings generated by FS algorithms after making a small perturbation to the training set. We mathematically justify the proposed measure based on the earlier and newly defined desirable properties. Additionally, we explore various heterogeneous ensemble techniques and compare them with traditional FS algorithms on real-world datasets. We perform extensive experiments to demonstrate that the heterogeneous ensemble techniques perform better than traditional FS algorithms with respect to the proposed measure and other performance metrics. [ABSTRACT FROM AUTHOR]

Cloud Computing provides a promising platform for executing large scale workflow applications with enormous computational resources to offer on-demand services. Tasks in a workflow may need different type of computing resources such as storage, compute and memory type. However, inappropriate selection of these resources may lead to higher makespan and resource wastage. In this paper, we propose an effective two-phase algorithm for provisioning of cloud resources for workflow applications by using its structural features to minimize makespan and resource wastage. The proposed approach considers the nature of the tasks which may be compute intensive, memory intensive or storage intensive. We assume a realistic cloud model similar to Amazon EC2 that provides virtual machines for different types of workloads. Most importantly, the workflow model used in our approach is assumed to contain limited information about the task which is applicable for real situation. The performance of the proposed work is measured using five benchmark scientific workflows. The simulation results show that the proposed approach outperforms two existing algorithms for all these workflows. [ABSTRACT FROM AUTHOR]

Due to enormous growth of Internet of Things (IoT) in the last decade, the amount of data generated through smart devices is increasing exponentially. Fog computing has emerged as a potential technology to deal such a huge volume of data in which task offloading is the most important aspect which has attracted significant attention. Many research works have been carried out, however, task offloading with latency sensitivity, reliability and result migration over a mobile user environment is still not widely addressed. In this paper, we propose a method for delay-sensitive and fault minimized task offloading for service requests made through a mobile/vehicular end user environment implemented via Software Defined Network (SDN) controllers integrated with the fog layer. This is a novel multi-phased model involving determining the optimal number of SDN controllers, clustering of the fog nodes (FNs) on the basis of SDN proximities, task prioritization and Gravitational Search Algorithm (GSA) based target FN selection. The simulation outcomes of our proposed approach show that there is a reduction in delay by around 23%–30% and around 60%–80% lesser number of tasks unassigned in each round as compared to two base algorithms. [ABSTRACT FROM AUTHOR]

Cloud computing has gained enormous popularity for on-demand services on a pay-per-use basis. However, a single data center may be limited in providing such services, particularly in the peak demand time as it may not have unlimited resource capacity. Therefore, multi-cloud environment has been introduced in which multiple clouds can be integrated together to provide a unified service in a collaborative fashion. However, task scheduling in such environment is much more challenging than that is used in the single cloud environment. In this paper, we propose three allocation-aware task scheduling algorithms for a multi-cloud environment. The algorithms are based on the traditional Min-Min and Max-Min algorithm and extended for multi-cloud environment. All the algorithms undergo three common phases, namely matching, allocating and scheduling to fit them in the multi-cloud environment. We perform extensive simulations on the proposed algorithms and test with various benchmark and synthetic datasets. We evaluate the performance of the proposed algorithms in terms of makespan, average cloud utilization and throughput and compare the results with the existing algorithms in such system. The comparison results clearly demonstrate the efficacy of the proposed algorithms. [ABSTRACT FROM AUTHOR]

• We propose an improved differential evolution (DE) algorithm by composite binary crossover and Pareto refining operations to the DE. • DE solves the quantifying fraudulent transactions problem by minimizing the selected feature cost while maximizing the quantification ability. • We show that the proposed DE outperforms four other variants of DE and five state-of-the-art EAs on various credit datasets. • The results are also statically validated through the Friedman test and Wilcoxon post-hoc analysis. [Display omitted] Identification of fraudulent credit card transactions is a complex problem mainly due to the following factors: 1) The relative behavior of customers and fraudsters may alter over time. 2) The ratio of legitimate to fraudulent transactions is highly imbalanced, and 3) Investigators examine a small segment of transactions in a reasonable time frame. Researchers have proposed various algorithms to identify potential fraud in a new incoming transaction. However, these approaches require significant human investigator effort and are sometimes misleading. To address this issue, this paper proposes an improved multiobjective differential evolution (DE) algorithm to estimate the distribution of fraudulent transactions in a set of new incoming transactions, referred to as quantifying fraudulent transactions. Our paper has three major novelties. First, we present the problem formulation of cost-based feature selection with maximum quantification ability. Second, we improve the DE by applying effective trial vector generation algorithms to the random control parameter settings to exploit the advantage of individual DE variants. Third, we develop the maximum-relevancy-minimum-redundancy-based Pareto refining operator to enhance the self-learning ability of individuals in Pareto solutions. We compare our approach against four other modifications of DE and five state-of-the-art evolutionary algorithms on real-time credit datasets in streaming and non-streaming frameworks using hyper-volume, two-set coverage, and spread performance metrics. [ABSTRACT FROM AUTHOR]

Data collection through mobile sink (MS) in wireless sensor networks (WSNs) is an effective solution to the hot-spot or sink-hole problem caused by multi-hop routing using the static sink. Rendezvous point (RP) based MS path design is a common and popular technique used in this regard. However, design of the optimal path is a well-known NP-hard problem. Therefore, an evolutionary approach like multi-objective particle swarm optimization (MOPSO) can prove to be a very promising and reasonable approach to solve the same. In this paper, we first present a Linear Programming formulation for the stated problem and then, propose an MOPSO-based algorithm to design an energy efficient trajectory for the MS. The algorithm is presented with an efficient particle encoding scheme and derivation of a proficient multi-objective fitness function. We use Pareto dominance in MOPSO for obtaining both local and global best guides for each particle. We carry out rigorous simulation experiments on the proposed algorithm and compare the results with two existing algorithms namely, tree cluster based data gathering algorithm (TCBDGA) and energy aware sink relocation (EASR). The results demonstrate that the proposed algorithm performs better than both of them in terms of various performance metrics. The results are also validated through the statistical test, analysis of variance (ANOVA) and its least significant difference (LSD) post hoc analysis. [ABSTRACT FROM AUTHOR]

Weighted moving average (WMA) is extensively used for short term forecasting to remove random fluctuations in cyclical data. However, it is susceptible to the number of previous observations (window size), used to predict the next element. A large value of window size increases the model risk whereas a smaller one poses parameter risk. For training data size of N, checking the window size ranging from 2 to N-1 needs O(N 3 ) time. In this paper, we propose MRF (Map Reduce based forecasting algorithm), which checks the possible window values in parallel and reduces the time complexity to O(N 2 ). The algorithm meticulously designs < key, value > for each window size, which distributes the job equally in the reduce phase. We have shown MRF to be the correct implementation of sequential algorithm. Extensive experiments were performed on MRF. The predicted values for real life meteorological and computing datasets generated using MRF are highly correlated to actual ones. Real and predicted are also validated using paired sample t tests. Results on synthetic data exhibits scalability of MRF algorithm. [ABSTRACT FROM AUTHOR]

Energy efficient data collection in a delay-bound application is a challenging issue for mobile sink-based wireless sensor networks. Many researchers have proposed the concept of rendezvous points (RPs) to design the path for the mobile sink. Rendezvous points are the locations in the network where the mobile sink halts and collects data from the nearby sensor nodes. However, the selection of RPs for the design of path has a significant impact on timely data collection from the network. In this paper, we propose an efficient algorithm for selection of the RPs for efficient design of mobile sink trajectory in delay-bound applications of wireless sensor networks. The algorithm is based on a virtual path and minimum spanning tree and shown to maximize network lifetime. We perform extensive simulations on the proposed algorithm and compare results with the existing algorithms to demonstrate the efficiency of the proposed algorithm of various performance metrics. [ABSTRACT FROM AUTHOR]

Cloud computing is one of the growing technology usage for the day-to-day business operations in today's IT industry. The diverse features of cloud such as on-demand self-service, quality of service, pay-per-usage pricing, virtualization and elasticity make the cloud more popular in industries as well as research communities. However, the mapping of the cloud resources in forms of virtual machines (VMs) to fulfill the customer requests is very challenging and a well-known NP-Complete problem. In this paper, we propose an allocation-aware task scheduling (ATS) algorithm for heterogeneous multi-cloud systems. The algorithm has three phases, namely matching, allocating and scheduling that aim to map the customer requests (or tasks) to the VMs of the clouds such that the overall completion time i.e., makespan is minimized. Moreover, the algorithm introduces a new phase called allocating to reschedule the tasks to meet the requirement of scheduling strategy. We perform rigorous experiments on benchmark as well as synthetic datasets and compare the experimental results by extending two existing multi-cloud scheduling algorithms as per the proposed model. The results clearly indicate that the proposed algorithm outperforms both the algorithms in terms of makespan and average cloud utilization. [ABSTRACT FROM AUTHOR]

Cloud Computing has grown exponentially in the business and research community over the last few years. It is now an emerging field and becomes more popular due to recent advances in virtualization technology. In Cloud Computing, various applications are submitted to the datacenters to obtain some services on pay-per-use basis. However, due to limited resources, some workloads are transferred to other data centers to handle peak client demands. Therefore, scheduling workloads in heterogeneous multi-cloud environment is a hot topic and very challenging due to heterogeneity of the cloud resources with varying capacities and functionalities. In this paper, we present three task scheduling algorithms, called MCC, MEMAX and CMMN for heterogeneous multi-cloud environment, which aim to minimize the makespan and maximize the average cloud utilization. The proposed MCC algorithm is a single-phase scheduling whereas rests are two-phase scheduling. We perform rigorous experiments on the proposed algorithms using various benchmark as well as synthetic datasets. Their performances are evaluated in terms of makespan and average cloud utilization and experimental results are compared with that of existing single-phase and two-phase scheduling algorithms to demonstrate the efficacy of the proposed algorithms. [ABSTRACT FROM AUTHOR]

Clustering is an efficient topology control method which balances the traffic load of the sensor nodes and improves the overall scalability and the life time of the wireless sensor networks (WSNs). However, in a cluster based WSN, the cluster heads (CHs) consume more energy due to extra work load of receiving the sensed data, data aggregation and transmission of aggregated data to the base station. Moreover, improper formation of clusters can make some CHs overloaded with high number of sensor nodes. This overload may lead to quick death of the CHs and thus partitions the network and thereby degrade the overall performance of the WSN. It is worthwhile to note that the computational complexity of finding optimum cluster for a large scale WSN is very high by a brute force approach. In this paper, we propose a novel differential evolution (DE) based clustering algorithm for WSNs to prolong lifetime of the network by preventing faster death of the highly loaded CHs. We incorporate a local improvement phase to the traditional DE for faster convergence and better performance of our proposed algorithm. We perform extensive simulation of the proposed algorithm. The experimental results demonstrate the efficiency of the proposed algorithm. [ABSTRACT FROM AUTHOR]

Abstract: In this paper, we propose, a novel DBSCAN method to cluster the gene expression data. The main problem of DBSCAN is its quadratic computational complexity. We resolve this drawback by using the prototypes produced from a squared error clustering method such as K-means. Then, the DBSCAN technique is applied efficiently using these prototypes. In our algorithm, during the iterations of DBSCAN, if a point from an uncovered prototype is assigned to a cluster, then all the other points of such prototype belongs to the same cluster. We have carried out excessive experiments on various two dimensional artificial and multi-dimensional biological data. The proposed technique is compared with few existing techniques. It is observed that proposed algorithm outperforms the existing methods. [Copyright &y& Elsevier]

Abstract: The Optical Multi-Trees (OMULT) is an interconnection network proposed by Sinha and Bandyopadhyay [B.P. Sinha, S. Bandyopadhyay, OMULT: An optical interconnection system for parallel computing, Lecture notes in Computer Science 3149 (2004) 302–312], for optoelectronic parallel computers. Various algorithms including matrix multiplication, DFT computation, sorting, prefix sum have been successfully mapped on this architecture. In this paper, we develop efficient parallel algorithms for some commonly used permutations namely, bit reversal, vector reversal, perfect shuffle, unshuffle and transpose on the OMULT network. Our algorithm for bit reversal permutation requires electronic moves +7 optical moves for data elements and electronic moves +3 optical moves for data elements; the vector reversal for data elements requires electronic moves +4 optical moves, where is the time for local vector reversal on data elements; the perfect shuffle for data elements requires () electronic moves +8 optical moves, where is the time for local perfect shuffle on data elements, and the transpose for data elements runs in at most three optical moves, all using – processors. [Copyright &y& Elsevier]

In recent years the multi-mesh network [Proceedings of the Ninth International Parallel Processing Symposium, Santa Barbara, CA, April 25–28, 1995, 17; IEEE Trans. on Comput. 68 (5) (1999) 536] has created a lot of interests among the researchers for its efficient topological properties. Several parallel algorithms for various trivial and nontrivial problems have been mapped on this network. However, because of its O(n) diameter, a large class of algorithms that involves frequent data broadcast in a row or in a column or between the diametrically opposite processors, requires O(n) time on an n×n multi-mesh. In search of faster algorithms, we introduce, in this paper, a new network topology, called multi-mesh of trees. This network is built around the multi-mesh network and the mesh of trees. As a result it can perform as efficiently as a multi-mesh network and also as efficiently as a mesh of trees. Several topological properties, including number of links, diameter, bisection width and decomposition are discussed. We present the parallel algorithms for finding sum of n4 elements and the n2-point Lagrange interpolation both in O(logn)1The base of the logarithm in this paper is assumed to be 2.1 time. The solution of n2-degree polynomial equation, n2-point DFT computation and sorting of n2 elements are all shown to run in O(logn) time too. The communication algorithms one-to-all, row broadcast and column broadcast are also described in O(logn) time. This can be compared with O(n) time algorithms on multi-mesh network for all these problems. [Copyright &y& Elsevier]

Sink mobility has been regarded as a widely accepted method for data collection in wireless sensor networks (WSNs) as it significantly improves network performance. Particularly, data collection using mobile sink based on rendezvous points (RPs) is a hot research topic which has been paid enormous attention to WSN community. However, efficient tour planning for the mobile sink (MS) is a challenging problem, especially for delay-harsh applications that require shorter paths of the MS. Existing literature finds this problem as NP-hard in nature, and thus nature-inspired algorithms are in demand as they can provide a near-optimal solution within acceptable time and space constraints. There are many schemes on MS tour planning that exist in the form of heuristics or nature-inspired algorithms; nevertheless, they leave out the scope for further research as most of them have not considered disjoint networks. Moreover, they fail to jointly optimize both the number of RPs and the number of MSs. To this end, this paper presents a novel scheme comprising of two algorithms based on the shark smell optimization (SSO) technique that solves the MS tour planning problem. The first algorithm is used to determine an optimal number of RPs and their locations. Based on this, the second algorithm optimizes the number of MSs so as to minimize the overall tour length. Each of the algorithms is developed with an efficient and novel particle encoding scheme along with the derivation of a fitness function. The tour planning is formulated as an Integer Linear Programming problem for the first algorithm and a Non-linear Programming problem for the second algorithm. Simulation results of our scheme confirm the improvement over the state-of-the-art algorithms. The results are also statistically validated through hypothesis testing using ANOVA and post hoc analysis. • An efficient and novel Shark Smell optimization-based scheme for data collection. • Joint optimization of set of RPs and their locations and number of mobile sinks. • Two novel particle encoding schemes and derivation of fitness functions. • Demonstration of superior performance of the scheme through extensive simulation. • Statistical validation of the results through ANOVA test and post hoc analysis. [ABSTRACT FROM AUTHOR]