Research Interests

My research interests mainly lie in the area of parallel and distributed computing, and the various applications. How to achieve efficiency, fault tolerance and security in a dynamic distributed environment is an interesting research topic to work on. My research interests include:

Research Projects

Formal Model for Cloud Computing

The cloud computing paradigm has recently received significant excitement and attention in the media. However, there is neither agreed exact definition of could computing, nor any formal model that we can rely on for analysis. We are currently developing a formal model that is able to capture the properties of cloud computing. Our work is proceeding in several aspects. 1) Efficiency: Develop an abstract model for Google's MapReduce parallel computing system and optimize its efficiency; 2) Fault Tolerance: Analyze the existing fault tolerance schemes and optimize agreement protocol, replica location and reexecution schedule; 3) Security: Address the security issues in such a large scale distributed environment.

I also maintain a list of related resources here.

Trust in Dynamic Multi-party Interactions

In order to deal with Byzantine behavior in multi-party interactions, we are currently building a trust and recommendation system, which adopts Bayesian inference to perform direct and indirect updates based on available evidence.

Auction in Multi-path Multi-hop Routing

To stimulate rational agents to behave truthfully in forwarding data packets, we model the multi-path multi-hop routing as a simultaneous-move one-shot game. By adapting the generalized second price auction mechanism, the proposed scheme results in Nash equilibria where the overpayment problem of VCG mechanism is greatly alleviated.

High Capacity Wireless Networking

We implement superposition coding technique on GNU software radio devices. By designing MAC/routing protocols that take advantage of packet mixing opportunity, the proposed scheme greatly improves network throughput based on our testbed experiments.

• High Throughput Routing with Superposition Coding and Successive Interference Cancellation. [PDF]
  Proceedings of the IEEE International Conference on Communications (ICC), 2011.

Link-aware Routing Metrics and Algorithms in Wireless Mesh Networks

Hop-count routing metric does not provide good performance estimation of routes in wireless networks. ETX, ETT and the extension WCETT are promising candidates for routing in mesh networks, however, none of them capture the mobility issues for the end-user mobile devices, such as smartphones, PDAs and laptops. We propose to incorporate link availability into the routing metric and protocol design such that the communication failure due to mobility could be minimized.

Reliable Multicast Routing in Mobile Ad Hoc Networks

To provide non-interrupted multicast connections, we propose to routing packets based on link availability metric and set up local alternative paths for vulnerable links. When the protected link fails, the pre-setup alternative path is used, thus improving the reliability of the whole multicast tree.

QoS, Fairness, Resource Allocation and Scheduling in Wireless Ad Hoc Networks

A number of research issues are involved in the design and implementation of protocols to operate ad hoc networks. One of the major issues is how to achieve fair bandwidth allocation. The objective of this project is to develop a bandwidth allocation and scheduling strategy for multi-hop flows in ad hoc networks. Based on graph theory, the proposed scheme decomposes the whole network traffic into maximal cliques, and charges prices based on cliques rather than links. The prices are feedback to source agents such that each of them could adjust the transmission rate dynamically according to network traffic and demand functions. Both in theory and practice, the proposed scheme achives max-min fairness.

Media Access Control in IEEE 802.11 (Undergraduate Thesis)

We focus on the IEEE 802.11 MAC protocol and analyze its performance in the aspects of fair access, conflict resolution and operation overhead. A dynamic algorithm is proposed to overcome the inefficiency of Binary Exponential Back-off algorithm in contention window adjustment.

Security System for a Campus Network (Undergraduate, Winner of ICM’04)

We construct an optimal defensive system for IT security for a university network. The modularized system employs Analytic Hierarchy Process (AHP) to clarify the miscellaneous risks and decompose the complex university network into nine simple subsystems. A fast search algorithm is used to determine a most suitable defensive technique for each subsystem. Based on this algorithm, we determine final polices and calculate the overall cost.

The Propagation Model of SARS (Undergraduate, Winner of CUMCM’03)

SARS is a serious infectious disease that could permanently damage human’s aspiratory system. We develop a mathematical model to analyze its propagation behavior. The prediction based on this model accords with the data collected in practice.