Avinash Srinivasan

PVFS: A Probabilistic Voting-based Filtering Scheme in Wireless Sensor Networks

In this paper, we study two important attacks in Wireless Sensor Networks (WSNs): the fabricated report with false votes attack; the false votes on real reports attack. Most of the existing works address the first attack while leaving an easy way for the attackers to launch the second. We draw our motivation from this and propose a Probabilistic Voting-based Filtering Scheme (PVFS) to deal with both of these attacks simultaneously. With the general en-route filtering scheme as the underlying model, PVFS combines cluster-based organisation, probabilistic key assignment, and voting methods to curtail these attacks. Through both analysis and simulation, we demonstrate that PVFS can achieve strong protection against both the aforementioned attacks while maintaining a sufficiently high filtering power.

DRBTS: Distributed Reputation-based Beacon Trust System

Wireless sensor networks (WSNs) have critical applications in diverse domains like environmental monitoring and military operations where accurate location of sensors is vital. One common method of location discovery uses a set of specialty nodes known as beacon nodes (BNs) that assist other sensor nodes (SNs) to determine their location. This paper proposes a novel reputation-based scheme called distributed reputation-based beacon trust system (DRBTS) for excluding malicious BNs that provide false location information. To the best of our knowledge, DRBTS is the first model to use the concept of reputation for excluding BNs. In DRBTS, every BN monitors its 1-hop neighborhood for misbehaving BNs and accordingly updates the reputation of the corresponding BN in the neighbor-reputation-table (NRT). The BNs then publish their NRT in their 1-hop neighborhood. BNs use this second-hand information published in NRT for updating the reputation of their neighbors after it qualifies a deviation test. On the other hand, the SNs use the NRT information to determine whether or not to use a given beacons location information, based on a simple majority voting scheme

TRACK: A Novel Connected Dominating Set based Sink Mobility Model for WSNs

The core functionality of a wireless sensor network (WSN) is to detect deviations in expected normal behavior and report it to the sink. In this paper, we propose TRACK - a novel sink mobility model exploiting the connected dominating set (CDS) property of a network graph. TRACK, to the best of our knowledge, is the first contemporary sink mobility model to exploit the CDS property for WSN lifetime longevity and secure data aggregation. In TRACK, the CDS of the given network is computed and then the minimum spanning tree (MST) of the CDS is constructed. Using the CDS-MST as the underlying framework, a Hamiltonian circuit (HC) is constructed, along which the sink is mobilized to traverse the network. Since TRACK, by the very definition of CDS, passes through the transmission range of every node in the network, data can be relayed directly from the source node to the sink, eliminating the need for multi-hop routing. By virtue of this property, nodes in the WSN are discharged from their routing obligations and data aggregation becomes more secure. Additionally, we propose an extended version of TRACK called M-TRACK in this paper. The extended model trades higher fractions of sensor energy with the objective of minimizing the length of sink trajectory. This consequently minimizes the delay between consecutive sink visits, mitigating buffer-overflow of sensors. M-TRACK does necessitate multi-hop routing, but keeps it within a bounded number of hops. We confirm the efficiency and robustness of our models via simulation and analysis, and show that our model can extend the WSN lifetime up to seven times that which can be achieved using a static sink.

iTrust : an integrated trust framework for wireless sensor networks

Designing security solutions for Wireless Sensor Networks is a challenging task due to the potential hostile and unattended environment in which they operate as well as their resource constrained nature. A trust management framework can be useful for detecting untrustworthy nodes under such operational conditions. In an unattended autonomous network, the attacker can capture a sensor node and modify its regular functioning. Consequently, the compromised node will thereafter behave erratically, which, in most cases, is observable by nodes in the corresponding neighborhood. In this paper, we propose iTrust- an integrated trust framework in which monitor nodes, a set of specialty nodes, will evaluate neighborhood nodes based on their behavior in a session wise manner. Monitor nodes, in promiscuous mode, will garner information about nodes in their neighborhood. After each session, they will share trust indices of each node with their neighbors, which is used for future decision-making. We have simulated iTrust framework with a tolerance of 5%--25% network error rate and evaluated its performance. We have further evaluated the attack detection effectiveness of iTrust framework by simulating different attack scenarios and confirmed its robustness to several known attacks.

A Novel CDS-Based Reputation Monitoring System for Wireless Sensor Networks

Reputation and Trust-based Monitoring Systems (RTMSs) have provided a ubiquitous framework for secure Wireless Sensor Network (WSN) computing. Employing sensors for neighborhood monitoring, which is secondary to their intended duties, depletes valuable and scarce resources which is counter-productive in WSNs. In this paper, we propose a novel, Connected Dominating Set (CDS)-based reputation monitoring system. Our model is the first attempt to employ a CDS-based monitoring backbone to securely aggregate the reputation of sensors without subjecting them to energy depletion or reputation pollution attacks encountered in existing reputation monitoring systems. Secure and certificateless node mobility and robustness to node replication and ID spoofing attacks are two vital by-products of our model. We confirm the performance of our model via simulation studies.

Secure and reliable broadcasting in wireless sensor networks using multi-parent trees

Summary Wireless sensor networks (WSNs) have been the focal point of research over the last several years. Broadcast communication is a key requirement for WSNs since many tasks in the network depend on broadcasting, including critical tasks like querying. Consequently, securing broadcast communication over sensor networks has become an important research challenge. Typically, broadcast communication involves two steps: broadcasting and acknowledging. In the broadcasting phase, the message is broadcast in the network. In the acknowledging phase, nodes that successfully received the broadcast message send an acknowledgment to the broadcast origination node, which in this paper is always the sink. The terms ‘sink’ and ‘base station (BS)’ are used interchangeably throughout this paper. Intuitively, broadcast communication has two important metrics: reliability and security. Though the reliability metric has drawn sufficient attention in the research community, the security metric has not. In this paper, we address both metrics with an emphasis on the former and address the Denial-of-Broadcast Message attacks (DoBM) in sensor networks. We propose a novel multi-parent tree-based model called the k-Parent Flooding Tree Model (k-FTM). We also present distributed algorithms for the construction of k-FTM and prove via simulation and analysis that the proposed k-FTM is robust against DoBM. Our Multi-Parent tree model enables the BS to detect DoBM very efficiently, even in the presence of a prudent adversary who focuses on remaining undetected by causing damage below the detection threshold. k-FTM is, to our best knowledge, the first fault-tolerant tree model that is both reliable and secure. Through simulations we confirm that our model achieves detection rates close to that of a static tree and a broadcast reliability close to that of blind flooding. Copyright

A Novel k-Parent Flooding Tree for Secure and Reliable Broadcasting in Sensor Networks

Securing broadcast communication over sensor networks has become an important research challenge. Intuitively, broadcast communication has two important metrics: reliability and security. Though the reliability metric has drawn sufficient attention in the research community, the security metric has not. In this paper, we address both metrics with an emphasis on the former and address the denial-of-broadcast message attacks (DoBM) in sensor networks. We propose a novel tree-based model called the k-parent flooding tree model (k-FTM) and present algorithms for the construction of k-FTM. The proposed k-FTM is robust against DoBM. It enables the base station to detect DoBM very efficiently, even in the presence of a prudent adversary who focuses on remaining undetected by causing damage below the detection threshold. k-FTM is, to our best knowledge, the first fault-tolerant tree model that is both reliable and secure. Through simulations we confirm that our model achieves detection rates close to that of a static tree and a broadcast reliability close to that of blind flooding.

Uncertainty Mitigation for Utility-Oriented Routing in Wireless Ad Hoc Networks

Link and node reliability are important metrics in wireless ad hoc networks. Therefore, evaluating and quantifying reliability has become the cornerstone of research in this field. Many existing wireless ad hoc network routing algorithms assume the availability of precise reliability information. This, however, is an unrealistic assumption given the dynamics of wireless ad hoc networks. Also, due to frequent changes in topology, reliability information is hard to collect, and oftentimes, inaccuracies can creep in. Therefore, a realistic method is needed to evaluate reliability by mitigating uncertainty in the estimation process. In this paper, we propose a novel reliability estimation model, account for uncertainty in the estimation, and design an uncertainty mitigation scheme. We then illustrate the effectiveness of our scheme in estimating reliability under various levels of uncertainty using a utility-oriented routing algorithm as a sample application. An extensive simulation study shows that the mitigation scheme significantly increases path stability and the long-term total benefit of the system.

SecLoc – secure localization in WSNs using CDS

Originally, the development of wireless sensor networks (WSNs) was motivated by military applications such as battlefield surveillance and land-mine detection. Over time, however, WSNs have found a wide range of applications in diverse domains such as industrial automation and monitoring, environmental and habitat monitoring, health-care applications, home automation, traffic regulation, smart hospitals, etc. In all these domains, the data sensed by the sensor nodes are reported to a central server called a base station, which then initiates appropriate actions based on the reported data. To this end, the location of sensors is very critical since the monitored event can be detrimental causing irreversible damage – such as forest fire, if the location of sensors is compromised and/or inaccurate. In this paper, we propose SecLoc – a novel localization method for WSNs, which can be easily extended to other wireless and mobile and ad-hoc networks. The proposed method exploits the connected dominating set (CDS) property of a network graph. SecLoc, to the best of our knowledge, is the first localization model to exploit the CDS property for accurate and secure node localization in WSNs. In out proposed method, a set of specialty nodes, called the beacon nodes, with large resource base, assume the role of Dominant nodes. The beacon nodes are responsible for both accurate and secure localization of nodes. We confirm the efficiency and robustness of our model through simulation results. Copyright

Duplicate File Names-A Novel Steganographic Data Hiding Technique

Data hiding has been an integral part of human society from the very early days dating back to BC. It has played its role for both good and bad purposes. First instances of data hiding dates back to 440 B.C. and has been cited in several works as one of the first known and recorded use of steganography. Several complicated Steganographic techniques have been proposed in the past decade to deceive the detection mechanisms. Steganalysis has also been one of the corner stones of research in the recent past to thwart such attempts of the adversary to subterfuge detection. In this paper we present a novel, simple, and easy to implement data hiding technique for hiding files with duplicate names. The proposed file hiding technique Duplicate File Names uses an innocuous file as the cover medium exploiting its name and reputation as a good file. This vulnerability was first discovered on a Windows 98 machine with DOS 6.1. We have tested this vulnerability on several different file systems to confirm that the vulnerability exists across file systems and not specific to older Windows file systems. Finally, we have discussed using this method for legitimate data hiding as well as detecting when employed for illegitimate data hiding.