Guest editorial special issue on signal processing in ... - IEEE Xplore

0 downloads 0 Views 144KB Size Report
Guest Editorial. Special Issue on Signal Processing in Networking. THE AREA of packet networking offers a wide range of problems for which signal processing ...


Guest Editorial Special Issue on Signal Processing in Networking


HE AREA of packet networking offers a wide range of problems for which signal processing can provide elegant and efficient solutions. This Special Issue is directed to and builds on the contributions of two research communities whose union is considerably larger than their intersection. On the one hand, researchers in signal processing have been drawn to interesting analytical problems related to modeling network traffic, performance, topology, and protocols. On the other hand, researchers in networking find themselves working on topics with signal processing aspects that have required novel types of data analysis, estimation, and prediction. This Special Issue lies on the interface between these two communities. We are delighted to be able to present to you, the reader, a timely special issue on signal processing in networking. From the signal processing point of view, the contents of our issue could be classified as follows. Most of the papers in this issue deal with estimation or detection of some fundamental quantity that affects quality of service, sensor localization, traffic distribution, or other parameter influencing network performance. For example, density estimation is considered in several of the papers on network tomography and sensor networks. Traffic prediction and anomaly detection also appear as the main problems in some papers. A couple of papers propose novel stochastic process models for Internet traffic. Several other papers offer novel mechanisms and protocols for both administered and ad hoc networks, including medium access control, collision resolution, and interference cancellation. The momentum for applying signal processing to problems in networking has been building for several years. The tremendous growth in networking hardware and software technology has permitted significantly higher throughput and user capacity in the Internet. However, the increasingly large and diverse user community’s demand for more bandwidth, mobility, and quality-of-service guarantees is also growing and will likely outstrip the capabilities of purely technological advances. This situation poses both a challenge and an opportunity for the signal processing community. We have skills in data analysis, modeling, and estimation that can generate novel solutions to diverse problems in the Internet, including topology discovery, traffic monitoring, and medium access control. Such skills are also applicable to next-generation wireless sensor and communication networks, including protocol design, interference cancellation, and controlling information flow in ad hoc networks. Papers in this Special Issue highlight recent advances that tackle some of these problems. The papers fall Digital Object Identifier 10.1109/TSP.2003.814459

into six complementary areas: Internet traffic modeling; on-line estimation and detection; network tomography; wireless communications; ad hoc networks; and sensor networks. Internet Traffic Modeling: Since the early 1990s, long-range dependence, impulsiveness, and fine-grain scaling phenomena have been found to characterize much of the observed Internet traffic. Internet traffic modeling has continued to be an important and attractive task to statistical signal processing researchers. Numerous stochastic models have already been proposed and new ones have emerged to capture increasingly fine-grain traffic behavior. Our issue brings two interesting contributions to this area. Online Estimation, Prediction, and Detection: Many of the traffic handling and control mechanisms suggested for Internet routers and traffic management functions at the edge require sophisticated, adaptable, and fast signal processing. In particular online schemes for tracking and predicting traffic patterns and for detecting changes in these patterns is becoming increasingly important to network administrators. Our special issue contains four papers that address these issues. Network Tomography: Partial knowledge of the network topology and performance will prove useful for various applications such as load balancing, verification of service, and detection of anomalous or malicious activities. Accurate knowledge of packet delay distribution and packet loss probability on individual links can be used for diverse applications such as service provisioning, service verification, topology discovery, network performance monitoring, and anomaly detection. This issue contains four papers in this thriving area. Wireless Communication Networks: Advances in wireless communications at the physical layer can be used to design more efficient higher network layer protocols. For example, in a centralized network, the performance of medium access control (MAC) protocol, which manages resource allocation to multiple clients, can be improved by exploiting physical layer techniques such as multiuser detection, interference cancellation, space-time coding, and channel estimation/equalization. Furthermore, in wireless ad hoc communications and sensor networks, the absence of centralized infrastructure makes well-designed signal processing at the physical layer indispensable. This issue collects four papers on signal processing in wireless access techniques, two on ad hoc networks, and two on sensor networks. This Special Issue contains a representative selection of some of today’s most active research in signal processing in networking. We hope the reader will find our sampling of activities in this multidisciplinary area of value. The editors

1053-587X/03$17.00 © 2003 IEEE



express their thanks to Arye Nehorai for soliciting this special issue and to Kevin Uherek for working tirelessly to ensure that it came to fruition. ALFRED O. HERO, III University of Michigan (e-mail: [email protected]) ILKKA NORROS VTT Technical Research Centre of Finland (e-mail: [email protected]) ATHINA PETROPULU Drexel University (e-mail: athina@ RUDOLF RIEDI Rice University (e-mail: [email protected])

Suggest Documents