Sophia G. Petridou

Towards realizable, low-cost broadcast systems for dynamic environments

A main design issue in a wireless data broadcasting system is to choose between push-based and pull-based logic: The former is used as a low-cost solution, while the latter is preferred when performance is of utmost importance. Therefore, the most significant advantage of a push system is the minimal cost. This fact implies that hardware limitations do exist in the case of push systems. As a consequence, every related proposed algorithm should primarily be cost-effective. This attribute, however, has been overlooked in related research. In this paper, popular broadcast scheduling approaches are tested from an implementation cost aspect, and the results render them only conditionally realizable. Moreover, a new, cost-effective, adaptivity oriented schedule constructor is proposed as a realistic, minimal-cost solution.

Cost-Aware Wireless Data Broadcasting

Research on push systems has naturally focused on improving the client serving time. However, in many cases the determinant factor for choosing between push and pull logic is the low central infrastructure cost. Adaptive push systems take this condition to the next step, requiring the central calculation of optimal broadcast schedules every few seconds or minutes, performed by relatively inexpensive server hardware. Aiming at introducing cost-aware wireless broadcasting systems, the Fast Optimization - Memory Conserving (FOMC) method is presented. The goal of the proposed method is to reduce the required computational power and memory of the central server, to the point that it can be implemented with mainstream hardware, and be incorporated to minute-scale adaptive systems. In order to demonstrate the importance of taking cost into account, FOMC is compared with the classical and influential Broadcast Disks method. While FOMC succeeds in requiring reasonable computational power and minimal memory, the Broadcast Disks method is rendered merely unrealizable in practice. Finally, in a first effort to minimize the scanning time for optimal broadcasting parameters, their relation with the total number broadcasted data items is being studied through a specially adapted multivariate illustration technique, and it is observed to be non-linear.

A divergence-oriented approach for web users clustering

Clustering web users based on their access patterns is a quite significant task in Web Usage Mining. Further to clustering it is important to evaluate the resulted clusters in order to choose the best clustering for a particular framework. This paper examines the usage of Kullback-Leibler divergence, an information theoretic distance, in conjuction with the k-means clustering algorithm. It compares KL-divergence with other well known distance measures (Euclidean, Standardized Euclidean and Manhattan) and evaluates clustering results using both objective function’s value and Davies-Bouldin index. Since it is imperative to assess whether the results of a clustering process are susceptible to noise, especially in noisy environments such as Web environment, our approach takes the impact of noise into account. The clusters obtained with KL approach seem to be superior to those obtained with the other distance measures in case our data have been corrupted by noise.

Survivability Analysis Using Probabilistic Model Checking: A Study on Wireless Sensor Networks

Survivability of a wireless sensor network (WSN) reflects the ability of the network to fulfill its mission despite the presence of abnormal events, such as failures. Given that sensor networks are receiving increasing attention due to the wide range of their applications, which include the critical areas of health, and military and security, survivability constitutes a key property for their study. This paper proposes a quantitative analysis for survivability evaluation of wireless sensors networks using probabilistic model checking. We define network survivability in line with four measures, namely, the frequency of failures, the data loss, the delay, and the compromised data due to a variety of failures. In particular, three types of failure events are considered, namely, node, link, and attack failures, which are due to power faults, communication faults, and black hole attacks, respectively. Then, we represent networks behavior with continuous-time Markov chains and randomly inject the aforementioned faults and attacks in the network to derive results that quantify the impact of them. Although the proposed study considers and provides results for a WSN architecture, it has the potential of being exploited in different networks with their own specifications.

Quantitative analysis for authentication of low-cost RFID tags

Formal analysis techniques are widely used today in order to verify and analyze communication protocols. In this work, we launch a quantitative analysis for the low-cost Radio Frequency Identification (RFID) protocol proposed by Song and Mitchell. The analysis exploits a Discrete-Time Markov Chain (DTMC) using the well-known PRISM model checker. We have managed to represent up to 100 RFID tags communicating with a reader and quantify each RFID session according to the protocols computation and transmission cost requirements. As a consequence, not only does the proposed analysis provide quantitative verification results, but also it constitutes a methodology for RFID designers who want to validate their products under specific cost requirements.

Towards energy consumption evaluation of the SSL handshake protocol in mobile communications

Secure Sockets Layer (SSL) protocol is the most popular security protocol, which provides confidentiality, authentication and integrity protection in Internet transactions. Nowadays, a large portion of them are executed by battery-powered handheld devices, such as personal digital assistants (PDAs), which are severely constrained by their resources, e.g. battery. This work launches an energy-aware study of the SSL Handshake protocol, the most complex part of SSL, considering a number of mobile SSL clients trying to simultaneously establish a secure connection with an SSL server. The proposed analysis is based upon the development of a Continuous Time Markov Chain (CTMC) model augmented with energy rewards, derived by the SSL execution on an iPAQ PDA. The novelty of our approach is that it constructs a model, tunable to communication-related parameters, such as the Bit Error Rate (BER), and verification errors, which overhead client-side energy consumption. In turn, both probabilistic and quantitative verification results are derived. The first ones validate the successful completion of handshakes and the later exhibit the impact of various parameters, such as cipher suites and verification error, on the energy requirements of SSL handshake in mobile communications.

Correlating Time-Related Data Sources with Co-clustering

A huge amount of data is circulated and collected every day on a regular time basis. Given a pair of such datasets, it might be possible to reveal hidden dependencies between them since the presence of the one dataset elements may influence the elements of the other dataset and vice versa. Furthermore, the impact of these relations may last during a period instead of the time point of their co-occurrence. Mining such relations under those assumptions is a challenging problem. In this paper, we study two time-related datasets whose elements are bilaterally affected over time. We employ a co-clustering approach to identify groups of similar elements on the basis of two distinct criteria: the direction and duration of their impact. The proposed approach is evaluated using time-related news and stocks market real datasets.

Nodes’ clustering in WDM star networks with real-time traffic

This paper proposes a novel scheduling scheme, namely node clustering with prioritized scheduling (NOC-PS), which is designed to handle real-time traffic in wavelength division multiplexing (WDM) star networks. NOC-PS is based on clustering techniques, while it takes into account the priority information of data packets. The clustering process aims at organizing the network nodes into groups in terms of their packetspsila requests per channel. Then, NOC-PS rearranges the nodespsila service order beginning from the cluster with long-length requests and ending to the cluster with short-length requests. The novelty of NOC-PS is that it applies separate clustering to nodes with high and low priority packets, while it also differentiates their scheduling, since high-priority packets have the privilege of being scheduled prior to low-priority ones. The simulation results indicate that the proposed scheme leads to a significantly higher throughput-delay performance for real-time traffic, without sacrificing the performance of non-real-time traffic.

An analytical approach to the design of wireless broadcast disks systems

The Broadcast Disks method is commonly used to schedule the data transmission in wireless push based networks. The performance of a system that utilizes this method depends upon several parameters. This paper presents the Optimization- Based Procedure (OBP), an analytical approach for estimating the optimal values of these parameters. The validity of the analysis is verified through simulation and comparison with other popular and traditional approaches. The new approach was found to be accurate and dominant in the vast majority of a wide set of test cases. It also provided a thorough insight of the systems stability and behavior in general.

An Efficient Clustering Oriented Algorithm for Message Scheduling on WDM Star Networks

Message sequencing and channel assignment are two important issues that have to be addressed when designing MAC protocols for optical Wavelength division multiplexing (WDM) networks. Up to now, popular approaches deal with channel assignment without however addressing the order in which the messages are scheduled. This paper presents a new reservation-based message scheduling algorithm for WDM star networks which is based on clustering techniques. The proposed clustering oriented-earliest available time scheduling (CO-EATS) creates groups of nodes whose messages are destined to common destination nodes. The goal of CO-EATS is to prevent consecutive messages from being destined to the same node. The simulation results have shown that the proposed scheme improves channel utilization and as a result it leads to higher network throughput while it keeps mean packet delay at low levels in comparison with conventional scheduling algorithms.