Srikanthan Thambipillai

Energy-efficient cluster-based scheme for failure management in sensor networks

Wireless sensor networks are characterised by dense deployment of energy constrained nodes. Owing to the deployment of large number of sensor nodes in uncontrolled hostile environments and unmonitored operation, it is common for the nodes to exhaust its energy and become inactive. The failing nodes create holes in the network topology causing connectivity loss, which may lead to critical information loss. To avoid degradation of performance, it is necessary that the failures are detected well in advance and appropriate measures are taken to sustain the network operation. An energy-efficient cluster-based technique is proposed to detect failures and recover the cluster structure. The proposed technique relies on the cluster members to detect the failures in the cluster and recover the connectivity. The proposed failure detection and recovery technique recovers the cluster structure in less than one-fourth of the time taken by the Gupta algorithm and is also proven to be 70% more energy-efficient than the same. The proposed cluster-based failure detection and recovery scheme proves to be an efficient and quick solution to robust and scalable sensor network for long and sustained operation.

A Cluster-Based Approach to Fault Detection and Recovery in Wireless Sensor Networks

Wireless sensor networks (WSN) are inherently fault- prone due to the shared wireless communication medium and harsh environments in which they are deployed. Energy is one of the most constraining factors and node failures due to crash and energy exhaustion are commonplace. In order to avoid degradation of service due to faults, it is necessary for the WSN to be able to detect faults early and initiate recovery actions. In this paper we propose a localized cluster based method for fault detection and network connectivity recovery which is energy efficient and responsive. Extensive simulations show that our proposed algorithm with cluster size 20 enables the network to recover from faults in one fourth of the time consumed by fault tolerant clustering method (Gupta algorithm). Our proposed method is also shown to consume less energy than crash faults identification method. The proposed algorithm lends well to the continuous operation and maintenance of robust sensor networks.

Size-restricted cluster formation and cluster maintenance technique for mobile ad hoc networks

Mobile ad hoc networks may be composed of a large number of nodes and hence a hierarchical cluster-based structure can be employed to address the scalability issues of the large network. In this paper we propose a size-restricted, distributed clustering strategy (cluster formation and cluster maintenance) for mobile ad hoc networks. A distributed approach where every node is responsible for the clustering decisions would avoid single-point bottleneck failures. We use a size restriction S, for the cluster formation and cluster maintenance. In addition, while forming the cluster we also use a diameter restriction K. The simulations show that our strategy gives rise to a lesser number of clusters when compared to other clustering algorithms proposed by Gerla et al., Fernadess et al. and Lin et al. This is attributed to the size restriction and cluster merging that we have incorporated. The size restriction helps in better management of resources inside the cluster. Our clustering technique uses a weight-based cluster head election strategy which results in less change in cluster head and higher cluster head lifetime when compared to the Least Cluster Head Changes (LCC) with lowest ID algorithm. The low change in cluster head and higher cluster head lifetime make our proposed algorithm suitable for high-mobility situations. The proposed clustering strategy could be used for building the clusters required in a hierarchical structure of the ad hoc network.

Vision-based patient monitoring: a comprehensive review of algorithms and technologies

Vision-based monitoring for assisted living is gaining increasing attention, especially in multi-modal monitoring systems owing to the several advantages of vision-based sensors. In this paper, a detailed survey of some of the important vision-based patient monitoring applications is presented, namely (a) fall detection (b) action and activity monitoring (c) sleep monitoring (d) respiration and apnea monitoring (e) epilepsy monitoring (f) vital signs monitoring and (g) facial expression monitoring. The challenges and state-of-art technologies in each of these applications is presented. This is the first work to present such a comprehensive survey with the focus on a set of seven most common applications pertaining to patient monitoring. Potential future directions are presented while also considering practical large scale deployment of vision-based systems in patient monitoring. One of the important conclusions drawn is that rather than applying generic algorithms, use of the application context of patient monitoring can be a useful way to develop novel techniques that are robust and yet cost-effective.

Unified Cordic Based Processor for Image Processing

The CORDIC algorithm, owing to its hardware efficiency, has been widely employed for computing complex mathematical operations. In this paper, a unified CORDIC-based processor for image processing is presented. For this, a wide range of popular algorithms like image enhancement in the spatial domain, frequency transforms, image rotation, edge detection etc. are chosen and their implementation using CORDIC is investigated. An integrated architecture that exploits the different operating modes of CORDIC is proposed employing an array of CORDIC-based processing elements. A qualitative analysis of the processor in terms of computation cycles and area complexity to process an entire image of size M × N is also presented.

DASCA: A Degree and Size based Clustering Approach for Wireless Sensor Networks

Sensor networks require scalable solutions to tackle huge number of sensor nodes. The scalable solutions have to be energy efficient and robust in order that reliability and network lifetime is increased. hierarchical (cluster based) models are known to be promising solutions to the problem of scalability. In this paper we propose a distributed clustering strategy which restricts the number of nodes in each cluster, S and also limits the number of next hop neighbours of a node in a cluster, D (admissible degree). The clustering algorithm is simulated and compared with two other clustering algorithms, HEED and LEACH. Simulation results depict that the energy drain in our algorithm is much lesser than the other two algorithms, thereby providing a longer lifetime when compared to the other algorithms. The proposed algorithm provides for almost 5 times increase in the lifetime of the network, when compared to the other two algorithms

A branch-and-bound algorithm for hardware/software partitioning

This paper investigates hardware/software (Hw/Sw) partitioning, a key problem in embedded co-design system. An efficient algorithm are proposed to optimally solve the problem in which the communication overhead is taken into account. The proposed algorithm constructs an efficient branch-and-bound approach to partition the hot path selected by path profiling techniques. The techniques for generation of good initial solution and the efficient lower bound for the feasible solution are customized in branch and bound search. Experimental results show that the partition result proposed by the new algorithm produces 10% increase in speedup as compared with the traditional approximate algorithm in most of the cases.

Knapsack Model and Algorithm for HW/SW Partitioning Problem

Hardware software partitioning is one of the most significant problem in embedded system design. The size of the total solution space for this problem is typically quite large. This problem has been investigated extensively. This paper is the first work to model the problem into a knapsack problem. We present a way to split the problem into standard 0-1 knapsack problems, so that most of the classical approaches for 0-1 knapsack problems can be directly applied. We use tight lower bound and tight upper bound on each of these knapsack problems to eliminate sub-problems, which are guaranteed not to give optimal results.

Estimating processor performance of library function

Fast estimation of processor performance can easily lead to large savings in time which would otherwise be spent in long time consuming execution on an instruction set simulator. The estimation of standard library functions are made difficult by the fact that standard libraries are sometimes delivered as binary files, which makes use of an instruction set simulator, the only option for performance estimation. In this paper, we present an estimation methodology which does not require execution of an application on an instruction set simulator. We use simple mathematical model of performance of standard library function which relates the number of instruction executed to the size/number of arguments passed.

A novel distributed cluster maintenance technique for high mobility ad-hoc networks

The dynamic nature of the mobile nodes in mobile ad-hoc networks (MANET), causes the mobile nodes to join and leave the clusters, thereby perturbing the membership of the clusters. Therefore cluster maintenance schemes are required to handle new admissions and releases of nodes in the clusters. Maintaining clusters in a distributed manner is more preferred in ad-hoc networks since it does not result in single point bottleneck failure. In a resource-limited network such as MANET (especially with nodes of varied capability), it is worthwhile to control the number of nodes per cluster for good planning and management of resources (such as memory, link bandwidth, battery power etc) in the cluster. In this paper, we propose a distributed cluster maintenance strategy that limits the number of nodes inside a cluster. However, we elect a cluster head for communication across the clusters. Our algorithm yields low number of clusters when compared to other algorithms taken for study. It is also observed that the number of cluster head changes is less and the life time of the cluster head is more. This behaviour of the cluster head is favourable to the high mobility environments because frequent updation of cluster head information and frequent re-election of cluster head may not be necessary.