Ravi Sankar

Time Series Prediction Using Support Vector Machines: A Survey

Time series prediction techniques have been used in many real-world applications such as financial market prediction, electric utility load forecasting , weather and environmental state prediction, and reliability forecasting. The underlying system models and time series data generating processes are generally complex for these applications and the models for these systems are usually not known a priori. Accurate and unbiased estimation of the time series data produced by these systems cannot always be achieved using well known linear techniques, and thus the estimation process requires more advanced time series prediction algorithms. This paper provides a survey of time series prediction applications using a novel machine learning approach: support vector machines (SVM). The underlying motivation for using SVMs is the ability of this methodology to accurately forecast time series data when the underlying system processes are typically nonlinear, non-stationary and not defined a-priori. SVMs have also been proven to outperform other non-linear techniques including neural-network based non-linear prediction techniques such as multi-layer perceptrons.The ultimate goal is to provide the reader with insight into the applications using SVM for time series prediction, to give a brief tutorial on SVMs for time series prediction, to outline some of the advantages and challenges in using SVMs for time series prediction, and to provide a source for the reader to locate books, technical journals, and other online SVM research resources.

A Survey of Intrusion Detection Systems in Wireless Sensor Networks

Wireless Sensor Networking is one of the most promising technologies that have applications ranging from health care to tactical military. Although Wireless Sensor Networks (WSNs) have appealing features (e.g., low installation cost, unattended network operation), due to the lack of a physical line of defense (i.e., there are no gateways or switches to monitor the information flow), the security of such networks is a big concern, especially for the applications where confidentiality has prime importance. Therefore, in order to operate WSNs in a secure way, any kind of intrusions should be detected before attackers can harm the network (i.e., sensor nodes) and/or information destination (i.e., data sink or base station). In this article, a survey of the state-of-the-art in Intrusion Detection Systems (IDSs) that are proposed for WSNs is presented. Firstly, detailed information about IDSs is provided. Secondly, a brief survey of IDSs proposed for Mobile Ad-Hoc Networks (MANETs) is presented and applicability of those systems to WSNs are discussed. Thirdly, IDSs proposed for WSNs are presented. This is followed by the analysis and comparison of each scheme along with their advantages and disadvantages. Finally, guidelines on IDSs that are potentially applicable to WSNs are provided. Our survey is concluded by highlighting open research issues in the field.

Signal processing methods for pulse oximetry

Current signal processing technology has driven many advances in almost every aspect of life, including medical applications. It follows that applying signal processing techniques to pulse oximetry could also provide major improvements. This research was designed to identify and implement one or more techniques that could improve pulse oximetry oxygen saturation (SpO2) measurements. The hypothesis was that frequency domain analysis could more easily extract the cardiac rate and amplitude of interest from the time domain signal. The focus was on the digital signal processing algorithms that had potential to improve pulse oximetry readings, and then test those algorithms. This was accomplished using the Fast Fourier Transform (FFT) and the Discrete Cosine Transform (DCT). The results indicate that the FFT and DCT computation of oxygen saturation were as accurate without averaging, as weighted moving average (WMA) algorithms currently being used, and directly indicate when erroneous calculations occur.

An Efficient Mutual Authentication and Access Control Scheme for Wireless Sensor Networks in Healthcare

Wireless sensor networks (WSNs) will play an active role in the 21th Century Healthcare IT to reduce the healthcare cost and improve the quality of care. The protection of data confidentiality and patient privacy are the most critical requirements for the ubiquitous use of WSNs in healthcare environments. This requires a secure and lightweight user authentication and access control. Symmetric key - based access control is not suitable for WSNs in healthcare due to dynamic network topology, mobility, and stringent resource constraints. In this paper, we propose a secure, lightweight public key - based security scheme, Mutual Authentication and Access Control based on Elliptic curve cryptography (MAACE). MAACE is a mutual authentication protocol where a healthcare professional can authenticate to an accessed node (a PDA or medical sensor) and vice versa. This is to ensure that medical data is not exposed to an unauthorized person. On the other hand, it ensures that medical data sent to healthcare professionals did not originate from a malicious node. MAACE is more scalable and requires less memory compared to symmetric key-based schemes. Furthermore, it is much more lightweight than other public key-based schemes. Security analysis and performance evaluation results are presented and compared to existing schemes to show advantages of the proposed scheme.

Impact of Realistic Mobility Models on Wireless Networks Performance

In order to assess the mobile ad hoc network (MANET) performance accurately, it is necessary that the mobility model used emulates closely the real life scenario. Random waypoint is generally used as the default mobility model in most network simulations, which exhibits movement patterns that are most non-humanlike. In this paper, we incorporate more realistic mobility models including entity mobility models (Manhattan grid and Gauss Markov) and group mobility models (reference point group mobility and column mobility) in a multi-path (Rayleigh) fading environment. The effects of realistic mobility characteristics like temporal and spatial dependencies of velocity and geographic restrictions on performance metrics such as energy-goodput, packet delivery ratio, and control overhead packets generated are studied in detail. The energy utilized to transmit, receive, and drop data and control packets are calculated for the various mobility models. This analysis helps in discerning the effect of mobility and fading on the performance of a MANET and further facilitates in determining the impact of mobility models on performance in various network scenarios

Handoff criteria for personal communication networks

The rapid growth in wireless personal communication networks (PCN) has placed demands on higher system capacity and performance. One way to increase the capacity of a cellular system is by reducing the size of the cells to microcells. But, in such microcellular environment the frequency of handoff also increases. Since the handoffs increase the load on the network system, they should be minimized. Generally hysteresis and averaging are used to decrease the number of handoffs. These have a deteriorating effect on the signal quality. It is important to have the right compromise between the number of handoffs and the overall received signal quality. An adaptive prediction based handoff algorithm has bee proposed to reduce the number of unnecessary handoffs. This is verified by a simulation of a microcellular CDMA/FDMA system.<<ETX>>

Two-Relay-Based Cooperative MAC Protocol for Wireless Ad hoc Networks

The cooperative communication approach promises improved throughput and delay performance by effective use of spatial diversity in wireless ad hoc networks. The CoopMAC I protocol proposed by Liu picks either a direct path or a relay path based on rate comparison to enhance average throughput and delay performances. However, its performance deteriorates under fading conditions due to lower direct path or relay path reliability. UtdMAC, which was proposed by Agarwal , performs better than CoopMAC I in terms of average throughput and delay performances due to improved transmission reliability provided by the backup relay path. Although it is better than CoopMAC I, UtdMAC does not fully benefit from higher throughput relay path (compared with the direct path) since it uses relay path only as a secondary backup path. In this paper, a new cooperative medium access control (MAC) protocol, which is termed the 2rcMAC protocol, is proposed for a small-sized network. The protocol makes use of two cooperating nodes to achieve superior throughput and delay performances, compared with the existing cooperative MAC protocols. The secondary relay path is invoked as a backup path for better transmission reliability and higher throughput through the relay path. Moreover, handshaking and single-bit feedbacks resolve contentions among relay nodes in proximity at the time and further provide the source node with rate information on source-to-destination, source-to-relay, and relay-to-destination links. Performance gains achieved by the 2rcMAC protocol under fast-fading conditions over the existing cooperative MAC protocols are compared and discussed. Simulation results clearly show an average throughput improvement of 7% and 25% and an average delay improvement of 94.8% and 98.9%, compared with UtdMAC and CoopMAC I, respectively.

Activity-oriented access control to ubiquitous hospital information and services

In hospital information systems, protecting the confidentiality of health information, whilst at the same time allowing authorized physicians to access it conveniently, is a crucial requirement. The need to deliver health information at the point-of-care is a primary factor to increase healthcare quality and cost efficiency. However, current systems require considerable coordination effort of hospital professionals to locate relevant documents to support a specific activity. This paper presents a flexible and dynamic access control model, Activity-Oriented Access Control (AOAC), which is based on user activity to authorize access permissions. A user is allowed to perform an activity if he/she holds a number of satisfactory attributes (i.e. roles, assignments, etc.) under a specified condition (e.g. time, location). Results of AOAC implementation in a realistic healthcare scenario have shown to meet two important requirements: protecting confidentiality of health information by denying an unauthorized access, and allowing physicians to conveniently browse medical data at the point-of-care. Furthermore, the average execution time was 0.078s which allows AOAC to work in real-time.

Audio Segmentation and Speaker Localization in Meeting Videos

Segmenting different individuals in a group meeting and their speech is an important first step for various tasks such as meeting transcription, automatic camera panning, multimedia retrieval and monologue detection. In this effort, given a meeting room video, we attempt to segment individual persons speech and localize them in the video, based on data from a single audio and video source. The segmentation method is driven by audio and enhanced by video cues. We used Bayesian information criterion (BIC) to segment the feature vector streams and graph spectral partitioning to cluster them. We compare our results with audio based segmentation method and our localization technique with the commonly used mutual information

An energy-efficient access control scheme for wireless sensor networks based on elliptic curve cryptography

For many mission-critical related wireless sensor network applications such as military and homeland security, users access restriction is necessary to be enforced by access control mechanisms for different access rights. Public key-based access control schemes are more attractive than symmetric-key based approaches due to high scalability, low memory requirement, easy key-addition/revocation for a new node, and no key pre-distribution requirement. Although Wang et al. recently introduced a promising access control scheme based on elliptic curve cryptography (ECC), it is still burdensome for sensors and has several security limitations (it does not provide mutual authentication and is strictly vulnerable to denial-of-service (DoS) attacks). This paper presents an energy-efficient access control scheme based on ECC to overcome these problems and more importantly to provide dominant energy-efficiency. Through analysis and simulation based evaluations, we show that the proposed scheme overcomes the security problems and has far better energy-efficiency compared to current scheme proposed by Wang et al.