Novelty detection is the identification of new or unknown data or signals that a machine learning system is not aware of during training. It is one of the fundamental requirements of a good classification or identification system since sometimes the test data contains information about objects that were not known at the time of training the model. This survey aims to provide an updated and structured overview of recent studies and approaches of novelty detection that have appeared in the machine learning and deep learning after the last review of Novelty detection in 2014. We surveyed on various techniques and applications of novelty detection in the main application domains.

Novelty Detection Review State of Art and Discussion of New Innovations in The Main Application Domains

Self-Organized Wireless Sensor Network (SOWSN) is a system of sensor nodes that takes global decisions through local interactions without involvement of any central entity. Wireless sensor nodes have constrained processing capability and energy. The key characteristic used to evaluate performance of Wireless Sensor Network (WSN) is its lifetime which depends on residual energy of nodes; hence the major challenge in WSN is the efficient use of available energy. Node clustering saves energy and also shows self-organization because global decision like Cluster Head (CH) selection is taken through mutual communication between nodes. In this paper, a new clustering method based on self-organization is implemented to boost lifetime of WSN. Sensor network is divided into regions. Cluster formation relies on Residual Energy (RE) and nearest Distance (D) from CH. Node with highest residual energy becomes CH. Rest nodes join the nearest CH. Clusters are broken when residual energy of CH falls below threshold energy; causing the sensor network to get self-organized into new clusters. RED also focuses to solve the energy hole problem caused due to higher energy consumption by CHs near Sink Node or Base Station (BS).

RED: Residual Energy and Distance Based Clustering to Avoid Energy Hole Problem in Self-organized Wireless Sensor Networks

In a distributed Wireless communication technology, sensor networks have been proposed for use in many applications such as environmental monitoring and surveillance. Due to the different computing power, sensing range and different ability, the deployment and topology in Heterogeneous WSN(HWSN) is complex compared to Homogeneous WSN. Clusters are formed in Heterogeneous Wireless Sensor Network and Cluster Head (CH) is selected based on the Initial Energy and Residual Energy with Low Draining Rate (RELDR). Data from the cluster members are sensed, aggregated and summarized by cluster head, processed between the clusters. A node in a network called Cluster Authority (CA) works as a supervising node that has remove list which collects the attacker information. A concept using Multiple Inputs and Multiple Outputs (MIMO) is implemented in the work which reduces the noise and improves the network performance. The algorithm detects the Blackhole nodes and removes from the network which is listed in the remove list of CA. The network is reconstructed using the legitimate nodes and cluster head selection is based on the Residual energy in the next rounds. Experimental results will be analyzed for the parameters Packet Delivery Ratio, Throughput, End to End Delay, Overhead and Energy. Performance analysis is done with the existing system using Single Input and Single Output(SISO).

Performance Analysis of Energy Efficient Cluster Based Heterogeneous Wireless Sensor Network against Malicious Attack

Wireless sensor network is an unionized structure that's formed via-massive wide variety of small lightweight densely deployed sensor nodes in hostile environment for monitoring tracking and detection and a sink node for gathering the sensed data for decision making process. As sensor nodes' batteries, having short duration of energy and can't be recharged. Hence, with the view to maintain network lifetime acceptably high, it emerge noticeably important to have specialized energy aware routing protocols to attain high scalability in such environments. Rather than direct and flat routing, hierarchical cluster based routing protocols make great extent contribution to increment the overall system scalability, robustness, lifetime, and energy efficiency, and also does load balancing and data aggregation with the intention to decrement the number of messages forwarded to the BS. This paper surveys a comprehensive review on recently discovered routing protocols proposed for clustering using various approaches for cluster head selection, for forming clusters and for routing the data, along with their advantages.

Recent advancements in clustering protocols for wireless sensor network

In this study, a new approach for novelty and anomaly detection, called HPFuzzNDA, is introduced. It is similar to the Possibilistic Fuzzy multi-class Novelty Detector (PFuzzND), which was originally developed for data streams. Both algorithms initially use a portion of labelled data from known classes to divide them into a given number of clusters, and then attempt to determine if the new instances, which may be unlabelled, belong to the known or novel classes or if they are anomalies, namely if they are extreme values that deviate from other observations, indicating noise or errors in measurement. However, for each class in HPFuzzNDA clusters are designed by using the new evolutionary algorithm NL-SHADE-RSP, the latter is a modification of the well-known L-SHADE approach. Additionally, the number of clusters for all classes is automatically adjusted in each step of HPFuzzNDA to improve its efficiency. The performance of the HPFuzzNDA approach was evaluated on a set of benchmark problems, specifically generated for novelty and anomaly detection. Experimental results demonstrated the workability and usefulness of the proposed approach as it was able to detect extensions of the known classes and to find new classes in addition to the anomalies. Moreover, numerical results showed that it outperformed PFuzzND. This was exhibited by the new mechanism proposed for cluster adjustments allowing HPFuzzNDA to achieve better classification accuracy in addition to better results in terms of macro F-score metric.

/pdf/a-hybrid-clustering-approach-based-on-fuzzy-logic-and-25agwswt.pdf

A Hybrid Clustering Approach Based on Fuzzy Logic and Evolutionary Computation for Anomaly Detection

Wireless sensor network (WSN) is collection of homogeneous or heterogeneous sensor nodes used to sense the data from physical environments and pass it to sink node. Data collection or sensing environment without any clustering mechanism is little hard which causes the high processing power, high battery consumption, maximize redundant bits etc. To avoid mentioned overheads, Distance based Angular Clustering Algorithm (DACA) with data aggregation techniques is used and to make sure the improvement, computation of various parameters are processed i.e. delay, throughput, network load, number of packets lost, PDR, jitter and packet efficiency. Our simulation results shows improved and energy efficient WSN.

Performance evaluation of Distance based Angular Clustering Algorithm (DACA) using data aggregation for heterogeneous WSN

Today the main concern is to improve the lifetime of Wireless Sensor Network (WSN) because WSN is collection of no. of small battery equipped sensor nodes which consume battery to communicate with each other and once battery is dead it is not possible to recharge it. They have the huge responsibility of sensing the environment, collect the data and then transfer it to Base station. For transmitting the data from edge node to parent node, sensor nodes must know the path which may or may not be the shortest that increases the overhead of path formation. To overcome the path formation overhead, clustering is one solution which further leads to the problem of non-coverage area. In this work, we are proposing a Distance based Angular Clustering Algorithm (DACA) on heterogeneous network which will overcome the path formation and non-coverage area problem by combining the data-aggregation algorithm to prolong the lifetime of wireless network.

Surinder Kaur

Papers

Performance evaluation of Distance based Angular Clustering Algorithm (DACA) using data aggregation for heterogeneous WSN

Distance based Angular Clustering Algorithm (DACA) for heterogeneous wireless sensor networks