Botnet attacks are a major concern in the cyber-security field. Cyber-attacks involving malware and bots are on the rise infecting more and more computers daily. While numerous approaches exist for detecting malicious traffic generated by bots, the ever-growing diversity of botnet categories continues to pose a significant challenge. As a result, a generalized approach is necessary. Differentiating between benign and malicious traffic largely depends on the traffic features that are being used to feed in the detection process. In this paper, we propose a new and better approach of selecting the best features for effective botnet detection. At first, we modify an existing metaheuristics-based algorithm, namely, binary particle swarm optimization (BPSO), apply it for feature selection and then compare its performance with existing feature selection algorithms. Moreover, we compared our modified BPSO algorithm’s performance with the classical version of it. Through experiments, we show the efficacy of our proposed feature selection algorithm. The results demonstrate that our modified BPSO algorithm outperforms state-of-the-art feature selection algorithms in multiple aspects. Additionally, it surpasses the classical BPSO variant, further highlighting its effectiveness and improvements.

Smile reveals distinctive facial features that makes a face more recognizable. However, little is known about the impact of smile on face recognition accuracy, especially in the context of automated face recognition systems using traditional machine learning algorithms. This is an important topic to study because understanding the influence of smile on face recognition could contribute to improve the performance of various applications that use face recognition underneath. In this paper, we propose a handful of novel features to capture the effect of smile on faces. With the new feature set, we then apply several machine learning algorithms on the data set that includes a comprehensive set of both smiling and neutral facial images of people with different gender, ethnicity and age groups. Our experimental result suggests that smiling faces produce high discriminating power that can be used effectively by the machine-learning algorithms to improve their accuracy and fairness.

Uncertainty quantification in a neural network is one of the most discussed topics for safety-critical applications. Though Neural Networks (NNs) have achieved state-of-the-art performance for many applications, they still provide unreliable point predictions, which lack information about uncertainty estimates. Among various methods to enable neural networks to estimate uncertainty, Monte Carlo (MC) dropout has gained much popularity in a short period due to its simplicity. In this study, we present a new version of the traditional dropout layer where we are able to fix the number of dropout configurations. As such, each layer can take and apply the new dropout layer in the MC method to quantify the uncertainty associated with NN predictions. We conduct experiments on both toy and realistic datasets and compare the results with the MC method using the traditional dropout layer. Performance analysis utilizing uncertainty evaluation metrics corroborates that our dropout layer offers better performance in most cases.

Precise image segmentation is one of the dominant factors in disease diagnosis. A typical application is the segmentation of breast ultrasound images, allowing radiologists to suggest what to do next. After emerging deep learning technology especially convolutional neural networks (CNNs), the image segmentation model achieved state-of-the-art performance in various medical applications such as cancer detection and classification, lung node segmentation, cell segmentation and so on. However, despite these successes, a big question arises: to what extent is the model certain about the predicted result? Generally, most deep learning models focus on high accuracy but not on uncertainty of predicted results, which is not enough to make a critical real-life decision such as a disease diagnosis, where a wrong decision can be life-threatening. Hence for making a crucial decision, it is essential that the predicted result will provide not only accuracy but also estimate model uncertainty. Our contribution to this research is to build a system that predicts pixel-wise semantic segmentation and provides uncertainty estimation of the predicted results. It is achieved by adding a dropout layer during training and using Monte Carlo dropout in inference. We evaluate our model with the breast ultrasound image dataset (BUSI) and compare the results with a few other state-of-the-art methods where our method outperforms others in terms of IoU.

By suppressing and cancelling the self-interference, the wireless full-duplex (FD) radios have shown promising results. Although theoretically the network throughput is assumed to be doubled in FD channels, the system throughput is far below 2 in reality due to the existence of hidden station problems. Therefore, a carefully designed MAC protocol is needed to avail full benefit of full-duplex radios. In this paper, we propose a novel FRTS/FCTS-based MAC protocol to further mitigate hidden station problems and improve the delay and throughput performance by introducing tertiary transmissions (TT) besides the previously known primary and secondary transmissions. Through realistic simulations in ns-3, we show that the proposed FD MAC protocol outperforms the conventional half-duplex MAC protocol for wireless networks in terms of system through-put. Also, the proposed FD MAC outperforms its counterpart busy tone-based FD MAC in terms of average delay.

A recent development emanating from RFID technology is Near Field Communication (NFC). Basically, NFC is a popular short range (<10 cm) wireless communication technology with applications in areas sensitive to security and privacy concerns such as contactless payment. Since NFC communications require very close proximity between two communicating devices (e.g., a smartcard and a terminal), it is generally believed that Man-in-the-Middle (MITM) attacks are practically infeasible here. Contrasting this belief, in this paper, we successfully establish MITM attack in NFC communications between a passive tag and an active terminal. We carefully present physical fundamentals of the attack, our engineering design, and results of successful attack implementation. Subsequently, we present the practical applicability of our MITM attack that exploits a potential vulnerability in EMV based contactless payment protocol, which arises due to separation between card authentication and transaction authorization phases. We demonstrate how an attacker can compromise the integrity of a contactless payment using a malicious MITM card, and also present multiple attack/victim scenarios to analyze different types of impacts of our attack. Further, we conduct rigorous experimental studies to analyze both hardware and practical ramifications of our attack. Finally, we propose a mechanism to detect the MITM attack based on experimental analysis that demands no additional hardware.

Sign language is a method of communication primarily used by the hearing impaired and mute persons. In this method, letters and words are expressed by hand gestures. In fingerspelling, meaningful words are constructed by signaling multiple letters in a sequence. In this paper, a system has been developed to detect fingerspelling in American Sign Language (ASL) and Bengali Sign Language (BdSL) using (data) gloves containing some suitably positioned sensors. The methodologies employed can be used even in resource-constrained environments. The system is capable of accurately detecting both static and dynamic symbols in the alphabets. The system shows a promising accuracy of (up to) 96%. Furthermore, this work presents a novel approach to perform a continuous assessment of symbols from a stream of run-time data.

Precise indoor localization is of great importance to automatically track people or objects indoors and plays a vital role in modern life. Despite a number of innovative research present in the literature indoor localization still remains an open problem. To trace the main reason we identify that in the present literature the tendency is to pinpoint the exact coordinates of a target device although most of the location based services (LBSs) do not require exact coordinates. To support LBS, one can simply divide the area of interest into several zones and perform ``zone-fencing'', i.e., find under which zone the user is currently located at. In this paper, we propose a zone-based indoor localization scheme using neural networks. With the results from real world indoor settings, we show that a number of empty clusters is generated when the traditional counter propagation network (CPN) is applied as is. But a slight modification to the CPN reduces the number of empty clusters significantly and provides promising accuracy. The proposed scheme outperforms ``k-Nearest Neighbor algorithm" (k-NN) and its promising accuracy makes it suitable for real-world deployment.

This work is meant to provide insights towards state-of-the-art Application Programming Interfaces (APIs) commercially available today for recognizing emotions from video footprint of facial expressions. We analyze and compare performance of four such integrable commercial APIs using standard Extended Cohn-Kanade Dataset (CK+) containing over 10, 000 images with facial emotions and some randomly collected real life images. We also discuss issues to understand their limitations and adaptation as well as enhancement strategies. For performance analysis, we introduce the performance metric moving average to find the primary expression in the displayed emotion of a video sequence. Finally, using the popular valence-arousal dimensions, we show how to (lexicographically) order six basic emotions anger, happiness, sadness, surprise, fear and disgust. By providing neutral performance comparison, we hope to fill the existing notable gap between researchers, solution providers and application developers working on emotion based user modeling.

The nodes in wireless ad-hoc networks frequently need to broadcast messages for route discovery and other services. The naive flooding requires every node to forward a message exactly once and causes serious performance bottleneck due to redundant traffic, contention, and collision.One convenient solution to overcome this problem is to construct a connected dominating set (CDS) as virtual backbone. Several centralized and distributed algorithms have been designed to compute CDS in order to reduce the number of rebroadcasts. However, none of the approaches aim at minimizing contentions. Contention is a phenomenon that occurs when a group of nodes want to transmit over a shared channel. During contention only one node gets access to the channel and the others defer their transmission for a later time. In this paper, a new heuristic dubbed as Contention Aware Connected Dominating Set (CACDS) has been proposed to construct a CDS that intelligently selects the nodes to reduce contention. A centralized algorithm has been deduced first to construct CDS using the global topology information. Since it is difficult to gather the entire topology information, a distributed counterpart has also been presented where a node selects a subset of nodes from its immediate neighbors as forwarding nodes based on 2-hop neighborhood information and reduces contention. Experimental results show that proposed heuristics significantly reduce contention and outperform other state-of-the-art algorithms in minimizing contention with a little increase in number of forwarding nodes.

Topology Control (TC) is a fundamental research problem where the goal is to determine a set of wireless links such that the composed topology satisfies some desirable properties. Some of the achievable properties are connectivity, symmetricity, planarity, minimal use of energy, sparseness, bounded maximum node degree etc. A topology is said to be planar if there is no link crossing in the topology. Planar topologies are heavily used by several protocols lying on different layers of the protocol stack. Several TC algorithms generating planar topologies have been proposed recently. Although sparseness is a key metric in evaluating performance of such topologies, none of the prior algorithms provide any mathematical model to determine sparseness of the topology. In this paper, we provide a generic analytical model for evaluating sparseness of planar topologies. The derived analytical expressions can be used in determining average node degree, topology size etc., without running simulations or prior to the deployment of real systems. The analytical expressions are validated through extensive simulation experiments.

Topology Control (TC) algorithms for multi-hop wireless networks create a connected communication subgraph that satisfies some topological properties by assigning appropriate transmission power to each node. A topology is said to be minimum-energy path-preserving if it preserves minimum energy paths between every pair of nodes. Creating minimum-energy path-preserving sparse topologies is a fundamental research problem in TC that has been addressed in several recent research works. Although sparseness is a key metric in comparing the performance of such algorithms, none of these prior works provides analytical models to determine the sparseness. In this paper, we provide a generic analytical model for evaluating sparseness of such topologies. The derived analytical expressions are useful in determining topology size without running simulations or prior to the deployment of real systems. Moreover, we demonstrate how to analytically couple sparseness of topologies with the radio transceiver parameters. The analytical expressions are validated through extensive simulation experiments.

Most of the topology control protocols for wireless ad hoc networks focus on minimizing power consumptions while maintaining connectivity. Those protocols are static in nature due to the fact that all the nodes use only a fixed set of (energy efficient) relays. However, using the same set of relays all the time causes some nodes to become heavily loaded creating congestion, unbalanced traffic and high latency in the network. Instead of generating a single topology with fixed relay sets, we propose a distributed algorithm that generates a family of connected topology with (possibly) disjoint relay sets. Thus, by creating different energy aware relay sets of each node the proposed algorithm performs a time-delayed rotation of data forwarding duty among those sets in order to evenly distribute loads in the network. We illustrate the efficiency and efficacy of the proposed algorithm through simulation experiments.

Internet based service providers, such as Amazon, Google, Yahoo etc, build their data centers (IDC) across multiple regions to provide reliable and low latency of services to clients. Ever-increasing service demand, complexity of services and growing client population cause enormous power consumptions by these IDCs incurring a major part of their running costs. Modern electric power grid provides a feasible way to dynamically and efficiently manage the electricity cost of distributed IDCs based on the Locational Marginal Pricing (LMP) policy. While recent works exploit LMP by electricity-price based geographic load distribution, the dynamic workload and high volatility of electricity prices induce highly volatile power demand and critical power peak problem. The benefit of cost minimization via geographic load distribution is counterbalanced with the high cost incurred by violating the peak power. In this paper, we study the dynamic control of electricity cost to provide low volatility in power demand and shaving of power peaks. To this end, a Model Predictive Control (MPC) electricity cost minimization problem is formulated based on a time-continuous differential model. The proposed solution minimizes electricity costs, provides low variation in power demand by penalizing the change in workload and alleviates the power peaks by tracking the available power budget. By providing extensive simulation results based on real-life electricity price traces we show the effectiveness of our approach.

Most topology control algorithms for wireless ad hoc networks strive to reduce energy consumption by creating a sparse topology with few long-distance links. However, in a sparse topology, the average path length is relatively large (increasing end-to-end delay), and the number of vertex-disjoint paths between source-destination pairs is relatively small (reducing fault-tolerance). Unlike traditional topology control algorithms that generate a single topology with a certain property, we propose a distributed algorithm that generates a family of topologies with a range of characteristics. The network designer can choose a suitable topology by simply tuning a single parameter Δ (power savings threshold), trading off energy savings for other features such as low latency and fault-tolerance. For the topologies generated by the proposed algorithm, we also provide an analytical model to estimate their structural density. The accuracy of the analytical model is validated with extensive simulation results.

Conventional broadcasting protocols suffer from network congestion, frequent message losses and corruption of broadcast messages due to a vast number of duplicate packets transmitting in the network. In this paper, we propose an efficient, scalable and reliable broadcast protocol to send a message in the entire network where every node is guaranteed to receive the message with low overhead and minimal cost. The scalability and reliability of broadcast transmission is ensured by composing the entire network in a hierarchy of grids/squares. A higher order square is made up of four lower order squares forming a quad tree architecture. With this grid architecture a node does not need to flood the broadcast packet to the entire network. Rather a node exploits the Geographic Forwarding mechanism to send the packet to a particular square. The rest of the work is done by the very first node receiving the packet destined for that particular grid. This node has now the responsibility to update its own grid with the broadcast packet. The whole procedure is repeated iteratively in every grid. Simulation results show that our proposed algorithm is reliable, scalable, robust and also outperforms some other promising broadcasting protocols that exist in the current literature.

Several protocols for ad hoc network try to reduce redundancy as an effective measure against broadcast problems. Though these protocols ensure good performance in a favorable environment, they perform poorly when node cooperation cannot be guaranteed due to intentional misbehavior or environmental hostility. As a result, the expected behavior of nodes to forward packets, which is the basic assumption of all broadcast approaches, cannot be achieved always. In this paper, we analyze the performance deterioration of these algorithms in hostile environment. As a remedy, we focus on the reverse direction and interestingly find that adding redundancy in a controlled manner can greatly compensate the performance loss due to node misbehavior. Here we propose a novel approach that tune the amount of redundancy so that reachability and routing load both remain at a satisfactory level. Comparing their relative performance we end up with the conclusion that though redundancy is undesired, controlled redundancy is effective in special situations like uncooperative environments.

IEEE 802.11 Medium Access Control (MAC) protocol employs two techniques for packet transmission; the basic access scheme and the RTS/CTS-based reservation scheme. A parameter called RTS Threshold determines which scheme to use. If packet size is smaller than the RTS Threshold then the basic scheme is used otherwise the reservation scheme is used. With current standard, the RTS Threshold is fixed. In this paper, we, first point out the advantages and disadvantages of RTS/CTS based scheme. Then we state the problems of having a fixed RTS Threshold. Next, we present a numerical way to fix the RTS Threshold adaptively based on network traffic. The proposed adaptive scheme creates a balance between the basic scheme and the RTS/CTS based scheme and optimizes the network throughput. Considering multi-hop networks with hidden node problems we have validated our proposal through simulation.

In a wireless ad hoc network, the main issue of a good broadcast protocol is to attain maximum reachability with minimal packet forwarding. Existing protocols address this issue by utilizing the knowledge of up to 2-hop neighbors to approximate an MCDS (minimum connected dominating set) via heuristics derived from techniques known as Self pruning and Dominant pruning. Our experiments show that, using these greedy choice heuristics result in a biased load distribution throughout the network. Some nodes become heavily loaded and consequently packets through those nodes, whether unicast or broadcast, experience significantly larger delay. Contention and collision also increase at some regions, while they are relatively low at other regions. In this paper we address these issues, and propose various methods to evenly distribute the load caused by broadcast packets. Our algorithms take various reactive measures to dynamically include less loaded nodes in the forward list, while maintaining total number of packet forwards low. Detailed simulation using ns-2 shows fair scheduling of resources and significant improvement in distribution of packet forwarding load, packet delay, latency and overall performance.

We discuss the issue of reliable transmission of bounded streams of data samples over simple wireless channels offered by low-end RF modules. Traditional solutions, i.e., window-based ARQ schemes, applied to such channels tend to waste bandwidth, because of the need to maintain duplex communication. Our case study shows how one can overcome the inherent limitations of low-bandwidth unstructured wireless channels and claim most of their capacity for transmitting useful data.

Many protocols for ad-hoc wireless networks perform poorly in situations when node cooperation cannot be enforced. This may happen because of the lack of global authority over the nodes, or because of the environmental hostility. The net outcome of such scenarios is what we call a selfish behavior of nodes, which may spontaneously refuse to forward some packets. While unicast communication has been studied in this context, little has been done to improve the reliability of broadcasting in the face of limited trustworthiness of nodes. In this paper, we propose two enhancements of the popular broadcasting algorithms based on dominant pruning, which significantly improve the reachability of nodes in an unreliable ad-hoc wireless network. The improvement can be viewed as an enhancement of the network's fault tolerance, as it applies to most scenarios in which a node may fail to carry out its forwarding duties.

The primary performance objective of a broadcast scheme in an ad-hoc wireless network is to reduce the total number of retransmissions needed to reach all nodes. Another (less appreciated) measure of interest is the broadcast latency, i.e., the amount of time required to complete the operation. We point out the tradeoff between the two measures and show how to significantly reduce the broadcast latency in networks whose MAC schemes are derivatives of IEEE 802.11.

We consider the power control problem in ad-hoc wireless networks. Local decisions regarding the transmission power level induce a subgraph of the maximum powered graph, G/sub max/, in which edges represent direct reachability at maximum power. We propose a new algorithm for constructing minimum-energy path-preserving subgraphs of G/sub max/. Its superiority over previous solutions demonstrates once again that strict protocol layering in wireless networks tends to be detrimental to performance.