Στο σεμινάριο Τηλεπικοινωνιών, Επεξεργασίας Σήματος και Δικτύων του Τμήματος Πληροφορικής και Τηλεπικοινωνιών του Πανεπιστημίου Αθηνών παρουσιάζονται ερευνητικές και άλλες συναφείς δραστηριότητες στον γενικότερο γνωστικό χώρο των τηλεπικοινωνιών, της επεξεργασίας σήματος και των δικτύων.
Οι ομιλίες παρουσιάζονται στην
Αίθουσα Τηλεδιασκέψεων Κέντρου Διαχείρισης Δικτύων. Πιθανές εξαιρέσεις θα
ανακοινώνονται ανά περίπτωση.
Η πλέον πρόσφατη παρουσίαση δεικνύεται με πράσινο χρώμα στο φόντο.
Η φυσική πρόσβαση στο Τμήμα Πληροφορικής & Tηλεπικοινωνιών, μπορεί να γίνει με μετρό-λεωφορείο (Metro-Buses).
Prof. Elias Aboutanios, University of New South Wales
Τρίτη 29 Ιουνίου 2010 (1:00 PM)
Prof. John G. Proakis, Department of Electrical and Computer Engineering, University of California San Diego
Τετάρτη 23 Ιουνίου 2010 (12:00 PM)
Prof. Arun Sen, Dept of Computer Science and Engineering, Arizona State University
Πέμπτη 10 Ιουνίου 2010 (11:30 AM)
Dr. Francky Catthoor, EE department of the K.U. Leuven
Τετάρτη 19 Μάιου 2010 (11:00 AM - 12:00 PM)
Dr. Michael Georgiopoulos, School of EECS at the University of Central Florida in Orlando, FL
Πέμπτη 8 Απριλίου 2010 (12:00 PM)
Dr. Konstantinos Stamatiou
Τετάρτη 16 Δεκεμβρίου 2009 (2:00 PM)
Dr. Spyridon Vassilaras
Assistant Professor, Athens Information Technology Center for Research and Graduate Education (AIT),
Τετάρτη 11 Νοέμβριος 2009 (15:30)
The detection and estimation of signals are areas of research activity that lie at the heart of the field of signal processing. Whatever the application is, the signal processing task most often reduces to detecting a signal of interest and then extracting the relevant information that is conveyed by the various parameters of this signal. In practice, these tasks are complicated by the presence of noise and other interfering signals. This talk will present the research into signal detection and parameter estimation and the major results achieved in four principal application areas: frequency estimation for satellite communications, space time adaptive processing for radar target detection, spectral estimation for nuclear magnetic resonance and bat acoustic calls, and antenna array processing for GPS signals.
Short CV: Elias Aboutanios graduated in 1997 from the University of New South Wales with a bachelor of Engineering, Electrical. During his undergraduate studies, he received a Co-op scholarship in 1994 followed by a Sydney Electricity scholarship in 1995. He worked at EnergyAustralia (formerly Sydney Electricity) between 1997 and 2001. In 1998 he was granted an Australian Postgraduate Award and enrolled in a PhD in Electrical Engineering at the University of Technology, Sydney. He was a member of the Cooperative Research Centre for Satellite Systems and pursued his research on frequency estimation for the tracking of FedSat – a low earth orbit satellite. From 2003 to 2007, he was a research fellow at the Institute for Digital Communications at the University of Edinburgh where he worked on radar target detection. Since 2007 he has been a senior lecturer at the University of New South Wales. His research interests are in signal detection and parameter estimation, and adaptive and statistical signal processing, with applications to radar, nuclear magnetic resonance spectroscopy, biological signal processing, and GPS signals.
This paper is focused on the characteristics of underwater acoustic communication channels and their implications on the design of bandwidth efficient modulation and demodulation techniques. Both noncoherent and phase coherent modulation and demodulation techniques are considered, and their performance, bandwidth efficiency and implementation complexity are compared.
Single carrier and multicarrier (OFDM) modulation and demodulation techniques are also considered. It is illustrated that the performance of high data rate single carrier systems is severely impacted by intersymbol interference (ISI) due to channel dispersion and that powerful equalization algorithms are required to reduce the performance loss caused by ISI. The use of coding in such single carrier systems makes it possible to perform turbo equalization in order to achieve further performance gains. Multicarrier (OFDM) modulation may be used to eliminate the need for high complexity equalizers in channels where the time variations are sufficiently slow relative to the symbol rate carried by each of the subcarriers in the multicarrier system.
Further increases in bandwidth efficiency are possible through spatial multiplexing. To achieve this objective, recent work on underwater acoustic communications has focused on MIMO techniques to increase the data rate and improve performance through signal diversity. This work on MIMO will be discussed.
As a final topic in this paper, we consider multiple access techniques suitable for use in underwater acoustic networks. The suitability of FDMA, TDMA and CDMA is assessed in view of the characteristics of underwater acoustic channels.
Short CV: Ο Καθηγητής John Proakis έλαβε το πτυχίο του από το Πανεπιστήμιο του Cincinnati, και τα πτυχία MSc και Phd. από τα Πανεπιστήμια MIT και Harvard, αντίστοιχα. Είναι επισκέπτης Καθηγητής στο Πανεπιστήμιο της California, San Diego και Επίτιμος Καθηγητής στο Πανεπιστήμιο Northeastern, USA. Ήταν Καθηγητής στο Πανεπιστήμιο Northeastern από το 1969-1998, όπου χρημάτισε πρόεδρος του τμήματος Ηλεκτρολόγων Μηχανικών και Associate Dean στη σχολή Μηχανικών. Τα ερευνητικά του ενδιαφέροντα εστιάζονται στις περιοχές των Επικοινωνιών και Επεξεργασίας Σήματος. Έχει συγγράψει οκτώ βιβλία. Τα καθένα από τα βιβλία Digital Communications, Digital Signal Processing, και Digital Processing of Speech Signals θεωρούνται ως η «βίβλος» στον αντίστοιχο τομέα. Έχει τιμηθεί με μία σειρά από σημαντικά βραβεία, όπως το ΙΕΕΕ Donald Field award, IEEE Signal Processing Education award, IEEE Education Society McGrawHill/Jacob Millman award, και τιμήθηκε με το σημαντικό βραβείο Athanasios Papoulis award της European Association for Signal Processing (EURASIP), που ήταν και το πρώτο που απονεμήθηκε μετά την ίδρυση του βραβείου.
Efforts are currently underway in the U.S. Air Force to utilize a
heterogeneous set of physical links (RF, Optical/Laser and SATCOM) to
interconnect a set of terrestrial, space and highly mobile airborne platforms
(satellites, aircrafts and Unmanned Aerial Vehicles (UAVs)) to form an
Airborne Network (AN). We propose an architecture for an Airborne Network to
provide a stable operating environment. We design algorithms to compute the
speed of movement of the airborne platforms, so that the resulting dynamic
topology remains connected at all times. Faults are likely to be localized in
military networks where an enemy attack may inflict massive but localized
damage to the network. To capture the notion of locality in fault tolerance
capability of such networks, we introduce the notion of region-based
connectivity. The attractive feature of the region-based connectivity as a
metric is that it can achieve the same level of fault-tolerance as the metric
connectivity, but with much lower transmission power for the nodes.
Optical bypass is an emergent technology that eliminates the need for optical-electrical-optical (O-E-O) conversion at most of the network nodes. However, the resulting network is still not all-optical, i.e., all connections cannot be established solely in the optical domain. Since the optical reach (the distance an optical signal can travel before its quality degrades to a level that necessitates regeneration) ranges from 500 to 2000 miles, regeneration of optical signals is essential to establish lightpaths of lengths greater than the optical reach. Given the optical reach of the signal, the goal of the regenerator placement problem is to find the minimum number of regenerators necessary in the network, so that every pair of nodes is able to establish a lightpath between them. We formulate the regenerator placement problem as a Connected Dominating Set problem in a Labeled Graph (LCDS) and provide a procedure for computing it. We evaluate the effectiveness of our approach on several networks.
Short CV: Arunabha (Arun) Sen received the Bachelor degree in electronics and telecommunication engineering from Jadavpur University, Kolkata, India, and the Ph.D. degree in Computer Science from the University of South Carolina, Columbia. He is currently a Professor in the Computer Science and Engineering Program at Arizona State University. He served as the Associate Chairman of the department responsible for Graduate Programs and Research from 2001-7. His research interest is in the area of resource optimization problems in wireless and optical networks. He primarily studies the algorithmic issues related to the problems in these domains and utilize graph theoretic and combinatorial optimization techniques to find solutions. He has published over 100 research papers in peer-reviewed journals and conferences on these topics. He has served many IEEE and ACM workshops and conferences either as a Program Committee member or as the Chair of the Program Committee. Currently, he serves as an Associate Editor of IEEE Transactions on Mobile Computing. His current research is sponsored by the U.S. Army Research Office, Air Force Office of Scientific Research, Defense Threat Reduction Agency and Office of Naval Research. He is also a member of the ASU team that won the prestigious DoD Minerva project award in 2009. He served as a Co-Chair of the First and Second International Workshop on Network Science for Communication Networks (NetSciCom 2009 and NetSciCom 2010) held in conjunction with IEEE Infocom in Rio de Janeiro in April 2009 and in San Diego in March 2010.
In software-intensive embedded systems, reliability is of prime importance. Complexity increase through increased feature integration, product life decrease, and a trend towards increasingly open systems press a need for better development methods to ensure continued reliable products. In this talk, we focus on techniques that improve the user-perceived functional reliability of high volume consumer electronic products, e.g. car audio systems. Our main objective is to minimize product failures that are exposed to the user. Our focus is on techniques in the context of Soft Error impact reduction for multi-media or wireless streaming applications. This mitigation is especially crucial when IC feature size is shrunk to levels where cosmic radiation will have a noticeable effect on the correct operation of embedded SoCs. To evaluate the results we have used a Transaction Level Model (TLM) simulator based on a grey-box task graph input. When we compare this with conventional error-correction code based mitigation schemes as our baseline, it shows that we can significantly reduce the hardware overhead while maintaining also the same reduced level of soft error rate impact observed at the system output. Using the developed techniques, we can nearly fully eliminate the conventional error correction area overhead of 12% or more in the larger memories, while keeping the performance overhead in the acceptable range. That is crucial for the cost-sensitive consumer electronics target domain. The test-vehicle used in this talk is derived from an industrial digital audio application for the automotive domain.
Short CV: Francky Catthoor received the engineering degree and a Ph.D. in electrical engineering from the Katholieke Universiteit Leuven, Belgium in 1982 and 1987 respectively. Between 1987 and 2000, he has headed several research domains in the area of high-level and system synthesis techniques and architectural methodologies, including related application and deep submicron technology aspects, all at the Inter-university Micro-Electronics Center (IMEC), Heverlee, Belgium. Currently he is an IMEC fellow. He is part-time full professor at the EE department of the K.U. Leuven. In 1986 he received the Young Scientist Award from the Marconi International Fellowship Council. He has been associate editor for several IEEE and ACM journals, like Trans. on VLSI Signal Processing, Trans. on Multimedia, and ACM TODAES. He was the program chair of several conferences including ISSS'97 and SIPS'01. He has been elected an IEEE Fellow in 2005.
ART (Adaptive Resonance Theory) was developed by Grossberg and its intent is to solve the perennial dilemma of β stability versus plasticityβ when a learning machine is designed. A number of ART neural network architectures have appeared in the literature, over the last 20 years, and have proven effective and efficient in solving a variety of classification problems. In this work, we present an approach to evolve ART Neural Network architectures (classifiers) that focuses on two distinct and sometimes competing objectives that of reducing the size of the neural network architecture, and that of reducing the generalization error. In particular, we propose the use of a multi-objective evolutionary approach to simultaneously evolve the weights and the topology of three well-known ART architectures; Fuzzy ARTMAP (FAM), Ellipsoidal ARTMAP (EAM) and Gaussian ARTMAP (GAM). We refer to the resulting architectures as MO-GFAM, MOGEAM, and MO-GGAM, and collectively as MO-GART. The major advantage of MO-GART is that it produces a number of solutions for the classification problem at hand that have different levels of merit (accuracy on unseen data (generalization) and size (number of categories created)). MO-GART is shown to be more elegant (does not require user intervention to define the network parameters), more effective (of better accuracy and smaller size), and more efficient (faster to produce the solution networks) than other ART neural network architectures that have appeared in the literature. Furthermore, MO-GART is shown to be competitive with other popular classifiers, such as CART (Classification and Regression Trees) and Support Vector Machines (SVMs).
Short CV: Michael Georgiopoulos received the Diploma in EE from the National Technical University in Athens, his MS degree and Ph.D. degree in EE from the University of Connecticut, Storrs, CT, in 1983 and 1986, respectively. He is currently a Professor in the School of EECS at the University of Central Florida in Orlando, FL. His research interests lie in the areas of Machine Learning and applications with special emphasis on neural network and neuroevolutionary algorithms, and their applications. He has published more than 60 journal papers and more than 180 conference papers in a variety of conference and journal venues. He has been an Associate Editor of the IEEE Transactions on Neural Networks from 2002-2006, and he is currently serving as an Associate Editor of the Neural Networks journal and the Computer Journal. He served as the General Chair of the S+SSPR 2008, the satellite workshop affiliated with the ICPR 2008 conference. He is currently serving as the Technical Co-Chair of the IJCNN 2011.
We consider a network where each route consists of a source, a number of dedicated relays and a destination at a finite distance, and the locations of the sources are determined according to a Poisson point process. Given a TDMA/ALOHA medium access control protocol, we evaluate the mean end-to-end delay and throughput in a typical route of the network, using a combination of tools from classical queueing theory and stochastic geometry. We then obtain the optimal number of relays and their positions such that the mean end-to-end delay is minimized, as a function of the network parameters and different packet arrival scenarios at the sources. Simulations in a realistic network environment demonstrate that the analysis provides useful routing guidelines in terms of reducing the packet end-to-end delay.
Short CV: Kostas Stamatiou received his Diploma in Electrical and Computer Engineering from the National Technical University of Athens in 1995 and his M.Sc. and Ph.D. degrees in Electrical Engineering in 2004 and 2009, respectively, from the University of California San Diego. His Ph.D. thesis focused on the performance analysis of interference-mitigating techniques in the context of cellular and ad hoc networks. His current research interests lie in the area of random networks and stochastic geometry.
Many theoretical results and performance optimization techniques in
wireless telecommunications and networks strive to maximize the system
throughput without taking into account the delay QoS requirements of
transmitted flows. In this talk, we will explore two such cases and
describe ways to optimize the system throughput under delay
The first case is Adaptive Modulation & Coding (AMC) for transmission of streaming media over wireless links. AMC aims at optimizing the trade-off between higher transmission rate and higher BER. When delay tolerant – loss intolerant data are transmitted, erroneously received packets can be retransmitted (using an ARQ protocol) as many times as needed until they get received without non-recoverable errors. In this scenario, traditional AMC algorithms which optimize the useful throughput of the wireless link offer an adequate solution. But for delay intolerant data flows (such as video streaming or VoIP) there is an upper limit to the number of retransmissions. Thus a packet is lost if it has not being correctly received within a certain period of time. In this talk, we use queuing models and Large Deviation techniques in order to derive AMC policies that minimize the overall probability of losing packets in both the no-retransmissions and persistent ARQ cases.
The second case is inspired by the well known work of M. Grossglauser and D. Tse which showed that mobility can drastically increase the capacity of ad hoc wireless networks. By using nodes as data carriers which physically transfer data closer to their intended destination before transmitting them over the air, the network throughput is considerably increased (compared to sending packets over multi-hop routes) at the expense of unbounded delays as the network size grows to infinity. The question addressed in this talk is “how can we design an algorithm that takes advantage of mobility to optimize the network throughput under a maximum allowable delay constraint?”. The problem is formulated and solved as a Dynamic Programming problem with some interesting twists to address the intermittent end-to-end connectivity case.
Short CV: Dr. Spyridon Vassilaras is an Assistant Professor
at the Athens Information Technology Center for Research and Graduate
Education (AIT). He received the Engineering Diploma in Electrical and
Computer Engineering from the National Technical University of Athens
in 1995 and both M.S. and Ph.D. degrees in Computer Engineering from
Boston University in 1997 and 2001 respectively. Prior to joining AIT
in November 2003, Dr. Vassilaras has also worked as a software
developer for ABB Industrial Systems in Sweden and as an application
developer / IT consultant for small Greek companies. His current
research interests include the performance analysis of
telecommunication networks using stochastic modelling, queuing theory,
Large Deviations theory, advanced simulation techniques (such us
variance reduction), linear and non-linear optimization. The prime goal
of this type of performance analysis is to provide Quality of Service
guarantees through cross-layer optimisation in fixed and wireless
networks. He is also interested in various aspects of wireless ad hoc
and sensor networks such as routing, transmission scheduling, cognitive
networking, network and data security (including node cooperation
issues), automotive telematics and wired/wireless integration. Dr.
Vassilaras is the author of a number of scientific papers published in
journals and presented in international conferences.