| Home |
| News |
| People |
| Projects |
| Publications |
| Workshops |
| Seminar |
| Software |
| Conferences |
| Funding |
| Collaborations |
| Resources |
| Directions |
 |
Current Projects |
||
| Pervasive Computing using Smart Phones |
Smart Phone is one device that holds great promise in realizing the pervasive computing vision. Goals of this project include: (1). Designing middleware and protocols for service provisioning on Smart Phones; (2). Designing middleware to support augmented reality applications on Smart Phones; (3). Characterizing energy usage of Smart Phone applications and developing energy optimization techniques. (4). Provide location based security.
| Networked Vehicular Systems |
The focus of our research in vehicular systems and networking is to build real systems in the area of Vehicular Ad-Hoc Networking. TrafficView is a traffic monitoring system that has been tested on real cars under real traffic conditions.
| FileWall |
The goal of our research in FileWall is to explore the benefits and limitations of both access monitoring and control policies for network file systems, using FileWall, a context-aware, programmable, network file system middlebox. Such context-aware analyses can be utilized to better understand and model user behavior and to perform context-aware network file system intrusion detection, non-intrusively. An additional goal of this project is to improve the programmability of the FileWall. The scope of this part includes providing efficient techniques for verification and validation within and across policies, as well as designing a graphical front-end.
| CovertFS |
The goal of this project is to create a web based covert file system, CovertFS, which facilitates secure file storage and sharing amongst a group of people and yet provides plausible deniability. The idea is to build the file system over a publicly available media service. Challenges are to map the local file system objects to the remotely hosted media in an efficient way such that covert traffic patterns appear as regular photo sharing traffic patterns.
| Authentication in Peer-to-peer Systems |
The increasing importance of peer-to-peer systems raises new challenges for authentication. These systems are open, span multiple administrative domains, and (potentially) contain malicious peers. These systems also have large numbers of unsophisticated users who expect automatic fault tolerant behavior out of the box.
Our research focus is to investigate and develop methods for achieving automatic fault tolerant authentication over open peer-to-peer networks.
Autonomous Transport Protocol (ATP) provides a reliable communication connection between two endpoints independent of their physical location. Autonomy allows dynamic endpoint relocation on different end hosts without disrupting the transport connection between them.
The Backdoor Architecture explores the use of non-conventional techniques in building systems that can perform failure detection and repair/recovery of state affected by a failure, while maintaining service to their clients. A remote healing system allows a remote machine to nonintrusively monitor a target machine and detect failures, then perform recovery and repair operations on it. Remote memory communication (RMC) is a powerful communication technique that provides interesting research opportunities for remote monitoring and intervention on a running system. In contrast with previous research that has used RMC mostly for its performance benefits, we take a novel approach on using it as a building block in the design of remote healing systems.
We envision that the use of embedded devices in cars will soon become a reality. To demonstrate the feasibility of the Smart Messages (SM) computing platform for real-world applications, we have developed EZCab, an application for locating and booking free cabs in densely crowded traffic environments (like Manhattan, where looking for a free cab can be quite an annoying experience).
Design and implement an architecture for a scalable federation of storage distributed across a cluster, which results in improved performance and availability of server applications in data center as well as wide-area environments. We call this architecture Federated File Systems (FedFS).
Migratory TCP (M-TCP) is a transport protocol compatible with TCP in which the client protocol stack can initiate migration of the remote endpoint of a live connection to an alternate server. Migration is transparent to the client application. M-TCP decouples the migration mechanism from migration policies that specify when should a connection migrate. Migration may be triggered according to some migration policy under conditions like server overload, network congestion, degradation in performance perceived by client, etc.
Design and build a Remote Services Platform (RSP) over RDMA interconnect technologies to provide a standard interface upon which various kernel services can be built. Using this platform, we are implementing in-kernel network storage services over RDMA using standard storage protocols (NFS and iSCSI).
In large computing clusters, loss of computation as a result of a failure becomes very costly. Our goal is to build a robust cluster computing environment that can efficiently (with low overhead during the failure-free execution) recover from failures and continue the distributed computation. Because Distributed Shared Memory (DSM) provides a familiar and easy to use programming environment for cluster computing, as a first step we chose to integrate fault tolerance in a DSM system.
The goal of the Service Continuations (SC) project is to provide an efficient OS-based solution for dynamic migration of client sessions established with an Internet service. Migration between geographically dispersed servers may be used according to various policies, to improve service quality as perceived by a client, to do load balancing at the server side, etc. Our goal is to provide an efficient and easy-to-use mechanism to support such policies.
Smart Messages project is to develop a computing model and a system architecture for networks of embedded systems (NES). The applications running over NES range from as simple as data collection and data dissemination in sensor networks to complex cooperative applications such as cars collaborating to adapt to traffic conditions or robots with intelligent cameras performing distributed object tracking.
Our work continues the research in Home-based Lazy Release Consistent (HLRC) DSM protocols started at Princeton University, focusing on scalability, fault-tolerance, adaptive protocols and non-scientific applications. We implement the shared memory abstraction as a software layer on top of a fast communication library. With this layer, a cluster of commodity PCs/workstations can provide the same programming interface as a hardware cache-coherent machine.
The goal of the Spatial Programming project is to design a simple programming model for networks of embedded systems that shields the programmers from the networking details. Similar to access to memory using variables, Spatial Programming provides network-transparent access to data and services distributed in the physical space using spatial references, a consistent naming scheme based on location and device properties. This project is joint work with EEL Lab.
Split-OS is a novel Operating System architecture for next generation servers. Over the past decade, devices like disk drives and NICs have been transforming into Intelligent Devices. Such devices include one or more processing engines, fast memory, and intelligent control programs (firmware). A system equipped with such devices can be thought of as a Cluster of Intelligent Devices. Our focus is to study the best strategies to split the OS functionality across these devices to improve performance and availability.
The goal of this project is to develop an architecture for the network subsystem that relies on offloading TCP/IP processing to dedicated processors, nodes or intelligent devices. This architecture should help to alleviate overheads resulting from network processing and also eliminate negative effects of co-location of server applications with computation intensive OS functions and network protocols.
Build User Level File Systems over RDMA interconnect technologies using standard storage protocols (DAFS and NFS), which provide low overhead access to network attached storage.
 |
 |