Production Number 03
Home | Research | Resume | Hobby | Links
 

My Research Projects at WiNG

I am a M.S. student in Wireless Networking Group (WiNG), which is lead by Professor Songwu Lu. UCLA WiNG group is engaged in research in wireless networking and mobile computing. My research interests are wireless services and systems, sensor networks, and network security. Here is a list of research projects that I have worked on during my Masters program.

DIRAC:  A Software-based Wireless Router for Mobile Computing Environment
GRAB:   GRAdient Broadcast Protocol for Large Scale Sensor Network
PEAS:    Probing Environment and Adaptive Sleeping Protocol for Large Scale Sensor Network
SCAN:   SeCurity for Ad-Hoc Networks
SCOPE: Distributed Packet Scheduling for Ad-Hoc Networks

DIRAC: A Software-based Wireless Router for Mobile Computing Environment

DIRAC is a software-based router system that is designed for wireless computing environment to facilitate the implementation and evaluation of channel-adaptive and mobility-aware protocols. The target application of DIRAC is a typical enterprise-wide packet-switched wireless data network. Our prototype implementation of DIRAC currently works with infrastructure mode of 802.11b; however, its design principles apply to other wireless technologies such as 802.11a, 802.11g, and wide-area 3G and 4G.
        DIRAC is a distributed router system that is consisted of a router core (RC) and several router agents (RA). The RC serves as the gateway router to serveral access points within the same subnet. A RA runs on a access point and it interacts with the RC through a communication protocol that is not specific to a particular wireless technology. RC gathers link-layer information from each RA and issues control commands based on the received information to be executed by specified RA(s). To demonstrate advantages and feasibility of the DIRAC design , we implemented three prototype wireless network services: link-layer assisted fast handover, channel-adaptive scheduling, and link-layer enforced policing. Our implementation and experiments show that our distributed wireless router provides a flexible framework, which enables advanced network-layer wireless services that are adaptive to channel conditions and host mobility.

Publications:

Links and Notes:

GRAB: GRAdient Broadcast Protocol for Large Scale Sensor Network

GRAB is a scalable, robust data delivery protocol that is designed for large scale sensor networks, which is deployed to monitor its environment. The targeting network is made up of thousands or even millions of small, low-cost, and unreliable sensors. We seek to design this protocol to provide robust data delivery using a large collection of unreliable sensors equiped with erroneous wireless links.
        To enable data forwarding between each sensor and the sensory data collection server (called sink), the sink builds and maintains a cost field by providing each sensor with the direction to forward incoming sensory data. Instead of forwarding the sensory data through a single forwarding path, each sensor forwards the data to a selected number of neighboring sensors, and thus the data traverse along an interleaved mesh from each source sensor to the receiving sink. GRAB protocol allows the source sensor to control the width of the forwarding mesh on-demand to provides various degree of delivery robustness. This mesh forwarding mechanism provides reliable, robust data delivery through the underlaying unreliable, erroneous sensor network. Our extensive simulation experiments shows that GRAB can successfully delivery over 90% of data with relatively low energy consumption despite up to 30% of sensor failures compounded with 15% packet losses resulting from wireless link error.

Publications:

PEAS: Probing Environment and Adaptive Sleeping Protocol for Large Scale Sensor Network

PEAS is a distributed energy consumption management protocol that is used to build long-lived, fault-tolerant sensor network using a large number of low-cost and unreliable sensors. This protocol targets densely deployed sensor network that is consisted of sensors with limited computational and battery resource. PEAS is designed to maintain an adequate number of working nodes to enable network connectivity with minimal amount of management overhead.
        PEAS extends the network life time by enabling only a necessary set of working sensor to maintain network connectivity and turning off redundant ones. PEAS is distributed protocol which does not require global coordination and clock synchronization. It maintains no per neighbor state and its operations are solely based on each sensor's local information. This distributed design requires low computation effort and minimal energy consumption from each sensor; therefore, PEAS is implementable on sensors with stringent amount of resource and it enables the network to be resilient to high sensor failures. Our simulations and analysis show that PEAS can adapt to very dense deployment with constant overhead, maintain adequate working nodes with up to 38% of failures, and extend a sensor network's functioning time in linear proportion to the deployed sensor population using less than 1% of total energy consumption.

Publications:

SCAN:   SeCurity for Ad-Hoc Networks

Publications:

SCOPE: Distributed Packet Scheduling for Ad-Hoc Networks

Publications:

  
    Production Number 03 Gary Zhong