Research Interests

         Computer Networks

         Collaborative Applications

         Peer-to-peer Systems

         Multimedia and Mobile Systems

         Network Security

         Applied Graph Theory

Completed Research Projects

Implementation and Empirical Evaluations of Floor Control Protocols on on PlanetLab

Nowadays users from different parts of the world participate in collaborative applications (online games, video-conferencing, distributed large-scale simulations) over the Internet. These applications require that at any point in time only one user can exclusively access a shared resource. The problem of providing exclusive access to shared resources in collaborative applications is known as the floor control problem. Centralized and distributed protocols for floor control problem have been proposed in the literature and simulation experiments have been conducted to study the performance of these protocols. None of these protocols have been tested on the real Internet. In this paper, we present empirical evaluations of different floor control protocols on PlanetLab network which is an overlay testbed that connects several academic institutions and industrial research labs all over the world in a virtual network.   We have implemented two flavors of distributed solutions the randomized Aloha and the scheduled DQDB (Distributed Queue Dual Bus) based protocols. We have also implemented the centralized protocol where the users send floor requests to a central node and the central node schedules the floor requests based on First in First Out. The protocols are implemented using Berkeley Software Distribution (BSD) sockets API. For the experiments, nodes from different PlanetLab sites are selected to resemble that users from different parts of the world are participating in a collaborative application. Average waiting time to gain a floor is used as the performance metric.

Students: Josh Dockery, Logan Daigle

Supported by: CF New Faculty Research Grant, CF Faculty Research Grant, CF Faculty Research Presentation Grant

 

GPRIDE A Generalized PRIority-based DistributEd Floor Control Protocol for Collaborative Applications

Internet users participate in different types of collaborative applications from different parts of the world. In these types of applications, users need to coordinate access to shared resources. This coordination is termed as floor control and researchers have proposed different types of floor control protocols in the literature. Certain types of collaborative applications require that a user with higher priority will be given preference over a user with lower priority at the time of floor assignment by the underlying protocol. This implies that the protocol needs to consider the priority of the user while assigning the floor. If two users are contending for the floor and their priorities are same, then the user who has requested the floor first will get the floor before the other user. These applications also require that this coordination should be performed without the help of any centralized controller. In this research, we have proposed GPRIDE - a Generalized PRIority-based DistributEd floor control protocol for collaborative applications. We have implemented our proposed protocol using BSD Sockets API and tested the performance of the protocol on the PlanetLab network for different conditions.

Student: Tao-Hsiang Chang

Supported by: CF Faculty Research Grant, CF Faculty Research Presentation Grant

 

M3DDVC: Multi-source Multicasting using Multi-cores with Delay and Delay Variation Constraints on Overlay Networks

 

Collaborative applications (online games, video-conferencing, distributed large-scale simulations) on overlay networks are continually growing in popularity. These applications require a multicasting subnetwork which has messages arriving at the destinations within a specified delay bound. They also require that all the destinations receive the message from the source at approximately the same time. The problem of finding such a multicasting subnetwork has been proved to be an NP-Complete problem and heuristics have been proposed in the literature. But these heuristics are designed for single source and multiple destinations. As most of the collaborative applications have multiple sources and multiple destinations, a shared multicasting subnetwork is required for this environment that will satisfy the delay and delay variation constraints. In this research, we have proposed a heuristic for designing a multicasting subnetwork (on an overlay network) with multiple sources and multiple destinations that will satisfy the end-to-end delay bound and achieve the tightest delay variation for each node in the multicasting group using multiple core nodes. The performance of the proposed heuristic is evaluated through extensive simulation experiments.

Student: Zachary C. Aardahl

Supported by: CF Faculty Research Grant

 

Load Balancing and Quality of Service Constrained Framework for Distributed Virtual Environments

 

Distributed Virtual Environments (DVE) have become increasingly popular over the last few years. Examples of DVEs are Massively Multiplayer Online Games (MMOGs), distributed interactive simulations, and shared virtual worlds. The service providers of DVEs need to ensure that certain Quality of Service (QoS) (messages delivered within a threshold delay) is guaranteed for the users participating in the DVE. In addition to ensuring QoS, the service providers want to balance the load on the servers that maintain the DVE. In this research, we have proposed a framework for DVEs which provides QoS to the users and balances the load among the servers. Our framework uses the concept of a virtual server which is a piece of software that does the processing for the DVEs. Each region in the DVE is maintained by an overlay of virtual servers. We have provided a heuristic that maps the virtual servers to physical servers, balances the load among the servers and ensures that the servers are not overloaded with objects. We have also presented a heuristic for creating a Degree and Diameter Bounded Multicast Tree of virtual servers for each region in the DVE which guarantees QoS for users in the DVE. We have conducted simulation experiments to evaluate the performance of our proposed framework.

Student: Noah Dietrich (College of Charleston)

Supported by: CF Faculty Research Presentation Grant

 

AD-NEMO: Adaptive Dynamic Network Expansion with Mobile rObots

Consider a situation in which a mobile wireless user needs connectivity to the Internet. One such situation arises in battlefield or at disaster recovery site where it may not be feasible to set up a fixed network. An alternative solution to this problem is to send out a team of mobile robots to establish and maintain the connection between the user and an established network that is connected to the Internet.  This includes the possibility of sending out new robots to join the team as needed, when the user moves beyond the range of the existing team, or if the signal becomes degraded for some reason. In this research we propose a framework called AD-NEMO (Adaptive Dynamic Network Expansion with Mobile rObots) where the mobile robots will create a dynamic and adaptive ad-hoc network to provide connectivity to the mobile user.  We have developed a prototype of our solution with actual hardware to study the feasibility of our solution.

Collaborator: Dr. George Rudolph

Students: Sean Feeney, Zachary C. Aardahl

Supported by: Dean Summer Research Grant, CF Faculty Research Grant

 

 

Current Research Projects

IMAIDS: Intelligent Mobile Agent-based Intrusion Detection System

Peer-to-peer network which is commonly known as P2P network is becoming popular nowadays. In this proposal we want to investigate intrusion detection mechanisms in P2P networks using intelligent mobile agents. An agent is a computing entity that performs some tasks or tasks on behalf of somebody. A mobile agent is an agent which is not bound to operate only in the system in which it started. An agent is intelligent when it can make its own decision based on the information it has gathered. Initially intelligent mobile agents will be placed in strategic locations (an algorithm will be designed to find these locations) in the P2P network to monitor the activities in the network to detect intrusions. These agents can roam around the network in order to identify distributed and sophisticated attacks. An algorithm will designed to find the roaming patterns of the agents. The agents should be able to stop the attacks once they are detected. A rule-based engine will be designed which will help the agents detect different types of attacks and take necessary actions to stop these attacks. In order to minimize the spread of an attack, the target hosts will be isolated from the rest of the network and other agents in the network will notified. An isolation and response mechanism will be designed so that the number of hosts affected by the attack is minimized. For performance evaluation the proposed IMAID (Intelligent Mobile Agent-based Intrusion Detection) system will be implemented on PlanetLab Network. PlanetLab is an overlay testbed network where several universities and research labs from different parts of the world are connected in a virtual network. The Citadel is a member of this network.

Collaborator: Dr. Muhammad Javed (Cameron University)

Students: Derek Bernsen, Hugh Urey

Supported by: CF Faculty Research Presentation Grant

 

Applying Data Mining to Intrusion Detection

We will look at different data mining techniques that can be applied to intrusion detection, and discover the techniques that will be best suitable for this purpose.  The goal is find and/or develop data mining techniques that will be able to efficiently, consuming the least amount of resources, scan a network log file, and detect maximum number of intrusions.  For this purpose, a survey of previously developed techniques will be performed, and in the absence of any suitable techniques, this project will be carried forward to develop a novel technique that can be applied in a multi-agent intrusion detection system.

Collaborator: Dr. Deepti Joshi

Student: Jonathan Ng