MARMALADeS (MARS)
This project investigates problems of designing and analyzing new efficient architectures and protocols for multicast and resource management/discovery in large-scale ad hoc and sensor (wireless mobile) networks.
MobiLib
In this project we proposed the TRACE framework to understand empirical mobile network user behaviors. Based on the Traces collected from a multitude of different environments, we construct Representations to Analyze various aspects of user behaviors (e.g., on-off usage pattern; skewed location visiting preferences; repetive association patterns, etc.). Through such analysis, important Characteristics of users can be identified, and Employed to design services and protocols. This environment-aware approach sheds lights on commonly used over-simplifying assumptions, leads to realistic user models, and helps to design more efficient protocols.
PUBLICATIONS
Mobility Modeling and Analysis
Mobility is one of the fundamental factors to influence the performance of wireless ad hoc networks. In this project we developed protocol-independent mobiltiy metrics and constructed the concept of mobility space - one needs a set of multiple mobility models that cover a wide range in each mobility metric to establish a thorough investigation. Mobility in turn determines the link duration and path duration, two key concepts in analyzing the communication reliability in ad hoc networks. Finally, we investigated how mobility interact with the protocols in ad hoc networks (for both path-based protocol (e.g., DSR) and encounter-based protocol (e.g., EASE)).
PUBLICATIONS
WLAN Performance Analysis
We design a generic model to capture WLAN (802.11 family) MAC protocol performances. The framework is flexible to accomodate different environmental parameters (e.g., number of stations, back-off parameters, etc.) and the mixture of stations running different extensions of 802.11 MAC protocol. Based on the generic model, a background traffic-aware rate-adaption protocol, BEWARE, is designed, analyzed, and currently under implementaion.
PUBLICATIONS
Wireless Security
As wireless network gains its popularity, a multitude of security-related issues are of concern in the new domain. We study these issues to reveal potential threats and provide solutions to them. Among the problems studied are (1) DDos attack in wireless networks and attacker trace back; (2) Impersonation attack and false data injection detection in sensor networks; (3) Worm (malicious code) propagation in sparse, encounter-based wireless networks and counter-worms; (4) Threats to user privacy and techniques to preserve it in large-scale user trace collection.
PUBLICATIONS
STRESS
The overall objective of STRESS is to provide tools and methodologies for systematic design and verification of networks to enable rapid deployment of robust network protocols and applications through automated test generation and their integration with simulation tools. An automatic topology and event sequence generation tool seeks to stress-test a given protocol and reveal potential pitfalls and performance anomalies. STRESS also aims to provide a more complete performance characterization of network and end-to-end protocols through sensitivity analysis.
PUBLICATIONS
ACQUIRE
Efficient query resolution and resource discovery is essential in large-scale sensor networks, and more so when the sensors are installed on moving objects, rendering the query resolution problem even more challenging. Flooding techniques, including global flooding or expanding ring search, may be very inefficient, especially in large wireless (spatial) networks and when queries are one-shot and frequent. The goals of the project are to develop and analyze a novel framework and protocols for scalable querying and resource discovery in next generation wireless sensor networks, using either wisely selected contacts or the natural gradient of sensed phenomenon to guide query forwarding.
PUBLICATIONS