Ijaz Haider Naqvi

Characteristics and classification of outlier detection techniques for wireless sensor networks in harsh environments: a survey

Wireless sensor networks (WSNs) have received considerable attention for multiple types of applications. In particular, outlier detection in WSNs has been an area of vast interest. Outlier detection becomes even more important for the applications involving harsh environments, however, it has not received extensive treatment in the literature. The identification of outliers in WSNs can be used for filtration of false data, find faulty nodes and discover events of interest. This paper presents a survey of the essential characteristics for the analysis of outlier detection techniques in harsh environments. These characteristics include, input data type, spatio-temporal and attribute correlations, user specified thresholds, outlier types(local and global), type of approach(distributed/centralized), outlier identification(event or error), outlier degree, outlier score, susceptibility to dynamic topology, non-stationarity and inhomogeneity. Moreover, the prioritization of various characteristics has been discussed for outlier detection techniques in harsh environments. The paper also gives a brief overview of the classification strategies for outlier detection techniques in WSNs and discusses the feasibility of various types of techniques for WSNs deployed in harsh environments.

Quarter-Sphere SVM: Attribute and Spatio-Temporal correlations based Outlier & Event Detection in wireless sensor networks

Support-Vector Machines (SVM) have received a great interest in the machine learning community since their introduction, especially in Outlier Detection in Wireless Sensor Networks (WSN). The Quarter-Sphere formulation of One-Class SVM (QS-SVM), extends the main SVM ideas from supervised to unsupervised learning algorithms. The QS-SVM formulation is based only on Spatio-Temporal correlations between the sensor nodes (hence the name Spatio-Temporal Quarter-Sphere SVM, ST-QS-SVM). Thus, it has a non-ideal performance. This work presents a new One-Class Quarter-Sphere SVM formulation based on the novel concept of Attribute Correlations between the sensor nodes, hence the name, Spatio-Temporal-Attribute Quarter-sphere SVM (STA-QS-SVM) formulation. Online and partially online approaches to Outlier Detection in WSNs have been presented using this formulation. The results indicate a significant increase in the Outlier Detection rates and a remarkable reduction in the False Positive rates over the previous formulation (ST-QS-SVM). The results of this novel technique also suggest that the partially online approach is as efficient as the online approach, thereby conserving significant computational and communication complexity. Moreover very high Event Detection rates have been reported for STA-QS-SVM, which have not been reported by ST-QS-SVM.

Support Vector Machine based fault detection & classification in smart grids

Smart Grids have recently attracted the attention of many profound research groups with their ability to create an automated and distributed energy level delivery. Computational Intelligence (CI) has been incorporated into various aspects of the smart grids, including fault detection and classification, which is a key issue in all the power systems. This paper presents two novel techniques for fault detection and classification in power Transmission Lines (TL). The proposed approaches are based on One-Class Quarter-Sphere Support Vector Machine (QSSVM). The first technique, Temporal-attribute QSSVM (TA-QSSVM), exploits the temporal and attribute correlations of the data measured in a TL for fault detection during the transient stage. The second technique is based on a novel One-Class SVM formulation, named as Attribute-QSSVM (A-QSSVM), that exploits attribute correlations only for automatic fault classification. The results indicate a detection and classification accuracy as high as 99%. Significant reduction (from O(n4) to O(n2)) in computational complexity is achieved as compared to the state-of-the-art techniques, which use Multi-Class SVM for fault classification. Moreover, unlike state-of-the-art techniques, both of these techniques are unsupervised and online and can be implemented on the existing monitoring infrastructure for online monitoring, fault detection and classification in power sytems.

One-class support vector machines: analysis of outlier detection for wireless sensor networks in harsh environments

Machine learning, like its various applications, has received a great interest in outlier detection in Wireless Sensor Networks. Support Vector Machines (SVM) are a special type of Machine learning techniques which are computationally inexpensive and provide a sparse solution. This work presents a detailed analysis of various formulations of one-class SVMs, like, hyper-plane, hyper-sphere, quarter-sphere and hyper-ellipsoidal. These formulations are used to separate the normal data from anomalous data. Various techniques based on these formulations have been analyzed in terms of a number of characteristics for harsh environments. These characteristics include input data type, spatio-temporal and attribute correlations, user specified thresholds, outlier types, outlier identification(event/error), outlier degree, susceptibility to dynamic topology, non-stationarity and inhomogeneity. A tabular description of improvement and feasibility of various techniques for deployment in the harsh environments has also been presented.

Frequency band selection and channel modeling for WNSN applications using simplenano

With the discovery of Graphene, different nano-scale components can be integrated to form a nano-sensor capable of transmitting in the tera-hertz range (0.1-10 THz). The transmitted signal gets attenuated when it passes through different types of molecules. Thus, channel models for wireless nano sensor networks (WNSN) are environment specific. For most of the applications, the nano-sensors are placed inside the human body, which may interact with other nano-sensors (inside the body) and other non nano-scale devices (outside the body). We present SimpleNano, a simple and novel channel model for WNSN applications in the tera-hertz range. SimpleNano is an approximation to log-distance path loss model along with the random (log-normally distributed) attenuation caused by the molecular absorption. For certain band windows within terahertz range SimpleNano experiences low molecular noise. These different band windows have been compared for a novel capacity performance metric i.e. spatial capacity per Hertz and the results prove that the proposed band windows achieve better throughput than the ones with larger molecular noise.

A DCCP Congestion Control Mechanism for Wired-cum-Wireless Environments

Existing transport protocols, be it TCP, SCTP or DCCP, do not provide an efficient congestion control mechanism for heterogeneous wired-cum-wireless networks. Solutions involving implicit loss discrimination schemes have been proposed but were never implemented. Appropriate mechanisms can dramatically improve bandwidth usage over the Internet, especially for multimedia transport based on partial reliability. In this paper we have implemented and evaluated a congestion control mechanism that implicitly discriminates congestion and wireless losses in the datagram congestion control protocol (DCCP) congestion control identification (CCID) framework. The new CCID was implemented as a NS-2 module. Comparisons were made with the TCP-like CCID and showed that the bandwidth utilization was improved by more than 30% and up to 50% in significant setups.

Real time energy efficient approach to Outlier & event detection in wireless sensor networks

The classification based schemes, especially One class Support Vector Machines (SVM) have received a great interest in Machine learning community based on their Outlier & Event Detection applications in Wireless Sensor Networks (WSNs). The quarter-sphere (QS) formulation of One-Class SVM, known as Spatio-temporal-attribute (STA-QS-SVM) has a high efficiency, but has a significant communication overhead. Since communication is much more expensive than computation is WSNs, therefore STA-QS-SVM is unsuitable for energy constrained WSNs. This work presents three partially online novel approaches based on STA-QS-SVM, which lead to a significant (upto 95%) reduction in the communication overheads. Most appropriate scheme of the three can be selected depending on the application. The results indicate that the proposed schemes give a performance comparable to that of STA-QS-SVM, with an added advantage of a significant reduction in communication cost. The results also suggest that the proposed schemes retain their performance even for very high fraction of outliers and events in the data sets. Thus the algorithms can be used for event detection in energy constrained WSNs deployed in harsh environments.

Radar ECCM against deception jamming: A novel approach using bi-static and mono-static radars

Bi-static radars have started to regain some recent attention over the past decade because of its abilities to counter deception jamming. This paper aims at utilizing the merits of bi-static radars that is used in combination with a mono-static radar. The proposed scheme counters the effects of range gate pull off (RGPO) and velocity gate pull off (VGPO) which are the two prime deception methods against modern radar systems. These methods are becoming ever increasing threats to ground based radars in modern era due to the advancements in digital radio frequency memory (DRFM). By using a combination of mono-static and bi-static radars in a simple yet effective configuration, it is shown that range deception and velocity deception jamming can be identified and the target can be localized.

Joint event detection & identification: A clustering based approach for Wireless Sensor Networks

Distributed clustering based techniques have been increasingly employed for outlier detection in Wireless Sensor Networks (WSNs). But despite its numerous advantages such as online and efficient computations and incorporation of spatiotemporal & attribute correlations, clustering has not been studied for event detection & identification, which is essential for smooth and reliable operations of large scale WSNs. This paper introduces the significance of clustering based event detection & identification to the research community. Further, it presents an online technique for joint event detection and identification that achieves a very high performance for synthetic and real data sets with a significant reduction in computational complexity as compared to the state-of-the-art techniques. A remarkable advantage of the proposed technique is that it can identify the key attributes in the ascending order of their contribution towards an event without incurring any additional complexity.

Robustness of a time reversal system: From ultra-wide to narrow bandwidths

The robustness of the time reversal (TR) scheme in a non stationary channel environment has been previously studied for ultra-wide band (UWB) channels. However, its robustness with different bandwidth specially for bandwidths significantly lower than ultra-wide bandwidths has not been established. In this paper, the performance of a TR communication system has been studied for large number of bandwidths for both UWB and narrow band systems in a non-stationary channel environment. The results show that TR scheme gives a robust performance for both wide and narrow bandwidths. The performance of the TR scheme also gets better for higher bandwidths. The analysis of the correlation of the channel and the TR responses for different bandwidths with respect to the responses having bandwidths that equal the coherence bandwidth of the channel has also been studied.