Mulugeta Libsie

Workload Characterization of Autonomic DBMSs Using Statistical and Data Mining Techniques

In this paper a model where an autonomic DBMS can identify and characterize the type of workload acting upon it is developed and the most important database status variables which are highly affected by changing workloads are identified. Two algorithms are selected for database workload classification: hierarchical clustering and classification & regression tree for classifying database workloads after running database workloads from TPC (Transaction Processing Performance Council) benchmark queries and transactions. The costs of these workloads are measured in terms of status variables of MySQL. A set of extensive experiments and analyses have been conducted and the results are presented in this paper.

The effect of node selfishness on the performance of WSN cluster-based routing algorithms

In Wireless Sensor Networks (WSNs), energy efficiency is among the most important factors since the lifetime of an unattended node is limited by its battery lifetime. In infrastructure-less WSNs, in order to use their resources efficiently, nodes can be organized hierarchically. In this case, long range communication is entrusted to a subset of nodes, the Cluster Heads (CHs): each CH acts as a gateway between the local nodes in its cluster and the other CHs in the network. When in this role, a node must stand a heavier traffic and consume more resources: fairness dictates that the role has to be taken in turn by each node in an area that fits the task. However, a selfish node may adopt several strategies to save its energy: it can make itself unavailable for the role of CH, or, when in that role, it can avoid relaying packets. In this work, we study the impact of selfish nodes on the performance of the cluster based algorithm LEACH. By taking into consideration several Quality of Service metrics, we quantify the extent of performance degradation as a function of the percentage of selfish nodes. We show that, when the percentage of selfish nodes is 75%, only 37% of packets reach the base station, network lifetime reduces to 40% of rounds and packet retransmission attempts is close to 63%.

A wireless sensor network framework for large-scale industrial water pollution monitoring

Design and implementation of Wireless Sensor Networks (WSNs) is a challenging task because WSN research is usually application specific and each application requirement brings with it a different set of constraints and design objectives. Effective use of WSNs in large scale and long-term monitoring of industrial water pollution requires satisfying two inherently contradicting requirements particular to this domain. On the one hand, it has to be time critical in order to provide early warning during compliance violations. On the other hand, it requires long-term data collection from a large number of distributed industrial sites for trend analysis for decision-making. We present the design of a comprehensive framework for using WSNs in industrial water pollution monitoring to assist environment authorities in decision-making regarding water pollution. The suitability and effectiveness of our proposed framework is verified with an experiment using the discrete-event simulator Castalia based on the OMNeT++ simulation platform.

Local Topology Aware Probabilistic Routing

Gossip-based packet forwarding is used in unstructured networks is to reduce traffic overhead in dense networks and to minimize early gossip termination in sparse networks. Unlike in flooding, where packets are forwarded to all the neighbors, in Gossip-based protocols packets are forwarded with some probability value p<1, to reduce redundancy. However this value has to be carefully tuned: if too small, early gossip termination is likely to occur, if too large, flooding storms can take place, as with the flooding protocol. In this work, we propose to use a forwarding probability based on local topology indicators, such as the effective node degree of the forwarding node: the choice of such probability takes into account the local topology. In a context where each node can have a different forwarding probability, another way of setting efficiently its value consists in further tuning such value for each message, based on the estimated level of completion of the corresponding communication task: to this purpose we propose to use a simple formula based on the messages hop-count. We validate these approaches by simulation using ns-2 in sparse and dense networks and show that they improve the performances in terms of traffic overhead and average end-to-end delay. In terms of packet delivery ratio, the proposed approach yields results comparable to those of the standard protocol AODV.

An evolutionary cluster-game approach for Wireless Sensor Networks in non-collaborative settings

Abstract Wireless Sensor Networks typically consist of a large number of sensor nodes with constrained resources. Cluster-based routing algorithms for WSNs try to preserve battery power by grouping nodes into multiple clusters: a single node in each cluster, the Cluster Head (CH), communicates with a Base Station on behalf of the others. In an ideal collaborative setting, sensor nodes should alternate in the role of CH. However, the cooperation of nodes is not granted in WSNs with more than one governing authority, where sensor nodes can behave selfishly, in order to save their own resources. In this paper, we propose a novel evolutionary cluster-head determination algorithm called GREET, based on an Evolutionary Game Theory (EGT) approach. In the proposed algorithm, individual nodes adapt their strategies on the basis of the outcomes of the interactions with other nodes and converge to an Evolutionary Stable Strategy (ESS) equilibrium. We show that this ESS corresponds to one of the desired behavioral outcomes. This outcome is obtained without the support of external cooperation enforcement mechanisms. In the study, we use an analytic model of the population evolution, based on the so-called replicator dynamics , as a guide in the choice of the mechanisms, then we adapt the approach to realistic more scenarios. We show, by means of a systematic simulation study, that the algorithm extends the network lifetime and provides a better packet throughput, w.r.t other standard WSN algorithms, such as LEACH and CROSS.

Collaborative packets forwarding to extend lifetime of multi-authority wireless sensor networks

In multi-authority Wireless Sensor Networks, governed by different authorities, resource constrained a node may be reluctant to forward packets received from other collaborating authorities, so as to save its own resources. If adopted by a significant percentage of nodes, such behavior can affect network performance and may reduce the service time of each node. To investigate such sensor nodes behavior, we used an Evolutionary Game Theory (EGT) approach. We studied a model of the system based on incentives and evolved the system according to standard replicator dynamics. We observed the emergence of a collaborative equilibrium. Based on this we designed a cluster based Collaborative Packets Forwarding (CPF) algorithm and showed by simulation that it is able to extend network lifetime.

Shape based customized node deployment approach for WSNs

The use of Wireless Sensor Networks (WSNs) has become an essential technology in many application domains including precision agriculture. Node deployment is a key issue in WSNs. In this paper, the nodes will be deployed over a network in a deterministic fashion in which the position of sensor nodes is known before deployment. These nodes are arranged regularly using triangular grid node deployment strategy over the given shape of the monitored region. This approach computes the minimum number of nodes needed to construct a sensor network and determine the position of sensor nodes using spatial coordinates with corresponding node deployment scheme and network topology. It uses real world deployment scenarios (i.e., taking the real shape of the monitored region of farmland) and determines the position of sensor nodes using spatial coordinates and this makes it more practical than others. This approach has to meet various requirements, such as coverage area, network lifetime, cost and ease of deployment, number of nodes, and adaptability. Moreover, we compared hierarchical routing algorithms called PEGASIS and LEACH to evaluate the network lifetime of the proposed scheme since the proposed node deployment scheme has an effect on routing. The results obtained from the implementation and evaluation of the proposed approach achieved the desired requirements and it is applicable to practical environments such as irregularly shaped farmland. We also show that node deployment is affected by the shape of the farmland. Thus, the required number of nodes and their positions can be adjusted adaptively to different shapes of farmlands.

A latency hiding framework for enhanced ubiquitous access to big data in a constrained digital ecosystem: application to digital medical archives

This paper presents our latency hiding framework for access to big data in a constrained digital ecosystem with application to digital medical archives. Aiming to enhance ubiquitous access of big data such as patient-oriented access of medical archives, we apply complex/multi-context prefetching to reduce latency thereby improving response time. We propose a formal model for prefetch requests rate and network workload or stress bound that takes into account a diverse set of constraints a digital ecosystem could be in. In addition to that, components of our latency hiding framework such as a generic multi-context functional architecture, use case model, medical database model with emphasis on API (abstracted patient information) and a high-level system architecture have been designed. The development of a complex or multi-context prefetch algorithm that uses a patients chief complaints, slackness sensitivity, popular content tag, user specified contexts and constraints is underway. A prototype system will also be developed to validate the proposed solutions. Moreover, input and output metrics will be developed to gauge the efficiency and effectiveness of the prefetch algorithm under development.