Sushanta Karmakar

A Neural Network based system for Intrusion Detection and attack classification

Anomaly based Intrusion Detection Systems (IDSs) are known to achieve high accuracy and detection rate. However, a significant computational overhead is incurred in training and deploying them. In this paper, we aim to address this issue by proposing a simple Artificial Neural Network (ANN) based IDS model. The proposed IDS model uses the feed forward and the back propagation algorithms along with various other optimization techniques to minimize the overall computational overhead, while at the same time maintain a high performance level. Experimental results on the benchmark NSL-KDD dataset shows that the performance (accuracy and detection rate) of the proposed ANN based IDS model is at par and in some cases even better than other IDS models. Owing to its high performance and low computational overhead, the proposed ANN based IDS model is a suitable candidate for real time deployment and intrusion detection analysis.

A Tree-Based Local Repairing Approach for Increasing Lifetime of Query Driven WSN

Query driven Broadcast through wireless sensor nodes also leads to the domain of event driven converge cast. A query-response based application in Wireless Sensor Networks(WSN) demands the correct delivery of data message at each sensor node. A Breadth-First Search(BFS) tree rooted at the base station offers shortest path traversal for each data message which utilizes the sensor resources efficiently. Resource constrained sensor nodes are highly prone to sudden crash. So the application demands a quick and smart approach to repair the tree when a node dies. In this paper a novel scheme has been proposed to locally repair the tree with constant round of message transmissions. Each node piggybacks a few bytes of extra information along with each query and response messages. Based on these piggybacked values each node calculates its alternate parent. When a parent node fails, its children can contact their respective alternate parents immediately to establish an alternate path to the root. Reduced communication cost in terms of extra message transmissions saves battery power at each node. Efficient query-response message handler ensures the correct delivery of messages. Fast repairing offers good Quality of Service(QoS). Simulation result shows that no message is lost except the one holding by the crashed node.

Adaptive broadcast by distributed protocol switching

Adaptation is a desirable requirement in a distributed system since it helps the system to perform gracefully under different scenarios. There are many adaptive algorithms for different problems. However the techniques are often application specific. In many distributed systems it may happen that the same problem has multiple protocols, each of which performs differently under different environments. In such cases adaptation can be achieved by dynamically switching between them as the environment changes. In this paper we illustrate the idea of designing adaptive distributed system using protocol switching by presenting an adaptive broadcast protocol that uses either a BFS tree or a DFS tree depending on the load of the system, both trees being rooted at the broadcast source. At low load a BFS tree is used as it reduces the broadcast delay (since the distance of any node from the root is always minimum in a BFS tree) while still keeping the load on any one node low. However at higher load a DFS tree is used to reduce the load on any one node since the degree of a node in a DFS tree is generally lower than that in a BFS tree. So the broadcast adapts to the network load by dynamically switching between a BFS tree and a DFS tree. In the proposed scheme, the switching is done by a middleware layered below the broadcast protocol. This separates the adaptation from the application and no change is needed in the application. We also ensure the application layer property that every broadcast message is correctly delivered to all the nodes, including messages sent when the switching is in progress. We assume the asynchronous model of distributed system with reliable but non-FIFO channels. Also it is assumed that no failure occurs during the switching. The message complexity of the algorithm that switches to a DFS tree is O(|E|) and is presented in the next section. The algorithm that switches to a BFS tree needs O(|V||E|) messages and is omitted in this paper due to lack of space.

Intrusion Detection Systems using Linear Discriminant Analysis and Logistic Regression

Anomaly based Intrusion Detection System (IDS) identifies intrusion by training itself to recognize acceptable behavior of the network. It then raises an alarm whenever any anomalous network behaviors outside the boundaries of its training sets are observed. However, anomaly based IDS are usually prone to high false positive rate due to difficulties involved in defining normal and abnormal network traffic patterns. In this paper, we employ two different statistical methods viz. Linear Discriminant Analysis (LDA) and Logistic Regression (LR) to develop new anomaly based IDS models. We then evaluate the performance of these IDS models on the benchmark NSL-KDD data set and analyze their performance against other IDS models based on Naive Bayes, C4.5 and Support Vector Machine (SVM). Experimental results show that the performance (Accuracy and Detection Rate) of both the LDA and the LR based models are at par and in some cases even better than other IDS models. Moreover, unlike the IDS model based on complex method like the SVM, the proposed LDA and LR based IDS models are computationally more efficient, which makes them more suited for deployment in real time network monitoring and intrusion detection analysis.

Fault resilience in sensor networks

Sensor network deployed for the critical infrastructure monitoring requires high degree of reliability in sensory data gathering, in spite of arbitrary node or sink failures. This paper proposes a robust data gathering scheme specially designed to provide guaranteed delivery of the sensory data for applications on the critical infrastructure monitoring. Redundancy in a sensor network, in terms of both the number of deployed sensors and the amount of duplicate data delivery, is explored to design an effective protocol that ensures the reliable data delivery while assuring the timeliness, connectivity and the sensing coverage. A set of active sensors is selected from all the sensors deployed, based on the network connectivity and the sensing coverage criteria that participates in the data forwarding process. Rest of the sensors go to the sleep state, and act as a replacement on the failure of an active sensor. The proposed protocol aims to find out multiple node-disjoint paths to multiple sinks, so that the loss of connectivity in one path due to node failure does not disrupt application services. The effectiveness of the proposed scheme has been analyzed using simulation results, and compared with other protocols proposed in the literature for reliable data delivery.

Convergecast tree management from arbitrary node failure in sensor network

Abstract The efficiency and reliability of convergecast in sensor network depends on the correct and efficient accumulation of data to the sink. A tree, rooted at the sink, that utilizes sensor resources properly is an ideal topology for data gathering. Resource constrained sensor nodes are highly prone to sudden crash. So the application demands a quick and efficient repairing of the tree when a node dies. The proposed scheme builds a data gathering tree rooted at the sink. The tree eventually becomes a Breadth First Search (BFS) tree where each node maintains the shortest hop-count to the root to reduce the routing delay. Each node collects some extra neighborhood information during the tree construction. Thus a little pre-processing at each node helps in taking prompt actions to repair the tree through local adjustment if any arbitrary single or multiple nodes fail in future. On failure of a node, each affected node in its vicinity fixes the parent through a pair of control message transmissions. Simulation results show that the repairing delay is significantly less in average and the convergecast messages are delivered with minimum data loss and no redundancy even in presence of node crash.

A reliable and total order tree based broadcast in wireless sensor network

A reliable broadcast demands that each data packet forwarded from the source node is received by all other nodes in the network without any loss. A flooding based Broadcast scheme incurs significant overhead due to explosion of packets in the network. So an efficient way of routing packets in order to broadcast saves the energy of resource constrained sensor nodes. The delivery of broadcast packets at each node in correct sequence is also crucial for certain applications. In this paper a tree based broadcast mechanism is proposed that constructs a spanning tree out of all nodes in the network rooted at the source of the broadcast. The broadcast tree eventually becomes a Breadth-First Search(BFS) tree where each node maintains the shortest path to the root. So packets routed through the shortest paths reduce transmission delay and packet loss. BFS tree based broadcast saves the battery power at each node by reducing the number of packet transmissions. The proposed scheme for broadcast is also crash tolerant. Each node computes a maintenance plan during the broadcast tree construction for future requirement. If a node dies, all the nodes in neighborhood repairs the tree locally within constant round of message transmissions. Each broadcast packet is assured to be delivered at each node without redundancy and maintaining the correct order.

Distributed deterministic 1–2 skip list for peer-to-peer system

Data management in the peer-to-peer system is a challenging task due to the random distribution of data among several participating peers. Efficient data structures like distributed hash tables (DHT) and its variants are designed and implemented to reduce the complexity of data management in such environment. However, DHT has its limitations in supporting range queries and its variants like distributed segment trees often perform poorly when the number of peers is high. Further, distributed lists and distributed balanced trees require significant amount of time for stabilizing after a new peer joins or a peer leaves. In this paper, a new distributed data structure called deterministic 1–2 skip list is introduced as an alternate solution for data management in the peer-to-peer systems. A deterministic skip list can be viewed as an alternate of a balanced tree, where the semantic locality of each key is preserved. Thus it can support the range queries as well as the single shot queries. This paper proposes three main operations on this data structure - searching data based on keys, insertion when a new peer joins, and deletion when a peer leaves. The correctness of the proposed operations are analyzed using theoretical arguments and mathematical proofs. The proposed scheme is simulated using NS-2.34 network simulator, and the efficiency of the scheme has been compared with DHT, DST, distributed list and distributed tree based data management.

Dynamic Tree Switching for Distributed Message-Passing Applications

The changes in environmental parameters may demand switching between underlying topologies for better performance of distributed message passing applications. Arbitrary topology switching using distributed tree construction may lead to loss or redundancy in delivery of application messages. In this work, a set of algorithms has been proposed for dynamic switching between two spanning trees to offer better adaptivity towards the environment for different applications. Here, two extreme cases of spanning trees, a Breadth First Search (BFS) tree and a Depth First Search (DFS) tree, rooted at the core node, have been considered for switching. The core node initiates the switching and all other nodes cooperatively change their parents on the fly maintaining the DFS or BFS properties as required. However, the application remains transparent to the switching that assures the availability of the system at any instance of time. Simulation results show that each application message is delivered correctly to the destination without any loss or redundancy. The proposed scheme is scalable and the control message overhead for switching is linear with respect to the number of edges in the communication graph. Furthermore, there is no control message overhead to assure the delivery of application messages at the time of switching.

Exploring gradient in sensor deployment pattern for data gathering with sleep based energy saving

The lifetime of sensor network depends on the efficient utilization of resource-constrained sensor nodes. Several MAC protocols like DMAC and its variants have been proposed to save critical sensor resources through sleep-wakeup scheduling over data gathering tree. For applications where data aggregation is not possible, the sleep duration decreases gradually from the leaves to the root of the data gathering tree. This results early failure of sensor nodes near the sink, and affects network connectivity and coverage. Deploying redundant sensors can solve this problem where a faulty node is replaced by a redundant node to maintain network connectivity and coverage. However, the amount of redundancy depends on the node failure pattern, and thus more number of redundant nodes required to be deployed near the sink. This paper proposes a gradient based sensor deployment scheme for energy-efficient data gathering exploring the trade-off among connectivity, coverage, fault-tolerance and redundancy. The density of deployment is estimated based on the distance of a node from the sink while dealing with connectivity, coverage and fault-tolerance. The effectiveness of the proposed scheme has been analyzed both theoretically and with the help of simulation.