ECE News


2004 Annual Report

Winter 2004 Connection

Fall 2003 Connection

2003 Annual Report

Head Letter

Chair Letter



Fuel Cells


Embedded Sys.

Wireless Video Networks

Demand for ECEs


ECE Research Update

Research News




Error &
Defect ID

in Hawaii

Cross Layer

UWB & Communications

Holey Fibers


Blue Diamond

Largest MOS
Controller Power Switch

New Power Monitoring

Virtual Hospital

Museum AVs

2001/2002 PhDs

2002 Patents

Winter 2003

Special Report:
What's Next for ECEs?
April 2003
Printer-friendly version (221Kb)

Thomas Hou has developed a novel, two-tiered architecture and traffic management plan for wireless video networks.The lower tier of small, low-cost video sensing nodes would capture video scenes, and an upper tier of more expensive group control nodes would collect and relay the data to the base station. His video networks scheme could be used for many applications, including military and homeland security surveillance and fire and wildlife monitoring.

Unlocking the Potential of Wireless Video Networks

Although wireless video surveillance or sensor networks are envisioned for monitoring wildlife, detecting natural disasters such as forest fires, for homeland security surveillance and sensing, or for real-time military information, they live still in the realm of future technology.

Thomas Hou, an expert in computer networks and telecommunications, has developed a novel architecture and traffic management plan to overcome some of the major obstacles in deploying wireless video networks. He recently won a prestigious Young Investigator Program (YIP) award from the Office of Naval Research (ONR) to develop this networking technology.

The ONR grants only 26 YIP awards nationwide each year to young faculty at U.S. universities whom it considers “the best and brightest young academic researchers.” The $300,000 awards support basic research in fields that are critically important to the technological superiority of the Navy and Marine Corps.

Real-Time Wireless Video

Current wireless sensor networks measure scalar data like temperature and pressure. Hou, however, envisions developing large scale, video-enabled wireless surveillance networks that can be deployed quickly and provide accurate, real-time visual data from the field. According to Hou, Naval use of the technology could include on-land and at-sea surveillance, video-assisted navigation, video assisted ship management, and remote monitoring of training exercises.

Wireless video sensor networks would be composed of interconnected, battery-powered miniature video cameras, each packaged with a low-power wireless transceiver that is capable of processing, sending, and receiving data.

Recent developments in device miniaturization, embedded computers, systems-on-a-chip, ultra-low-power streaming video technology, and wireless communications will soon provide the components, Hou explained. “Now we need to develop the networking technologies to handle the huge traffic volume of real-time video and have such networks operate for us as long as possible under limited battery power.”

Architecture, Traffic, and Lifetime Issues

Hou is working on three major issues, network architecture, scalable traffic management, and network lifetime. “How can we design the network so that it can grow as needed and not be limited in size? How do we manage the very heavy video traffic so that all the information gets where it needs to at the right time? How do we maximize the lifetime of such network based on components with limited battery power?”

According to Hou, traditional wireless sensor networks are based on a flat, homogenous architecture in which every sensor has the same physical capabilities and can only interact with neighboring sensors in the network. Such networks simply cannot handle the amount of traffic generated by video applications, he explained. “The amount of processing required on each node in terms of computing and communications and the power required to operate it would make such a network unfeasible to implement in reality,” he said.

Hou has developed a novel two-tiered wireless network architecture and a traffic management technique to solve the issues of large scale. His approach would employ a lower tier of small, low cost video sensing nodes (VSN) for capturing video scenes, and an upper tier of more expensive group control nodes (GCN), which would collect and relay data between the VSNs and the base station.

Two-Tiered Network

The sensing nodes would be equipped with a video capture camera and a transmitter/receiver device and would be placed in strategic surveillance locations. The VSNs would be low-power devices capable of sending information to nearby VSNs in their group and to their group control node (GCN). The control nodes would be physically different from the VSNs, containing the processing power to aggregate data in their groups, and transmit aggregate data over large distances. While the VSNs could be disposable, the GCNs would be recyclable.

Scalable Traffic Management

The two-tiered architecture will not solve all the scalability issues. “There will still be very large amounts of data to be managed at the upper-tier GCN level, particularly as all the traffic aggregates as it nears the base station. This is compounded by video applications, where each packet within a video flow must meet stringent delay bounds. If a packet of information for a video frame arrives too late, it’s useless,” he explained.

Typically, to achieve a guaranteed (or an acceptable) quality of service (QoS), network nodes, or routers, must maintain information on the rate and delay guarantees for each packet in a traffic flow and use that information to schedule the packets. “A traffic management system that requires this QoS state information simply is not scalable for a large-scale wireless video surveillance network supporting thousands or even tens of thousands of flows,” he said. “If the network size grows or the number of video flows exceeds a GCN’s memory size, the quality of service cannot be guaranteed.”

Hou plans to apply a technique he helped develop for wired networks, called the virtual time packet scheduling, that transforms the state information for each flow into smaller sets of information that can be encoded in each packet’s header. With this technique, the GCN is no longer required to maintain per-flow state information, and packet scheduling can be done solely based on the new state information encoded in the packet header.

The Wireless Medium Challenge

The challenge is that, although the virtual time packet scheduling solves the scalability issues for the upper-tier GCN nodes, the GCNs communicate through a shared wireless medium, Hou said. “In a wireline network, each node has dedicated links that can be used for transmitting and receiving packets without interfering with other nodes,” he said. “We need to solve the issues relating to media access control, or sharing the wireless medium. There are many alternatives, such as TDMA, CDMA and random access, each with its own possibilities. We do not want to introduce excessive delays due to media contention.”

Prolonging Network Life

Hou’s third major challenge in deploying large-scale wireless video sensor networks is determining the useful life of a large network that is constructed with thousands of devices carrying limited battery power. “The severe energy constraint on each node adds a new venue in networking research, particularly in flow routing. Not only do we need to be able to design a network that can operate, but we need to design algorithms for network flow routing such that network lifetime can be maximized. The traditional smallest cost path approach may not give maximum network lifetime. We must reconsider some basic design issues in network flow routing so that the network lifetime can be prolonged as much as possible,” he said.

“Developing good solutions for these networking problems is the key to unlocking the full potential of a large-scale wireless video sensor network,” Hou added.

Privacy Statement | Contact Webmaster

© 2006 Virginia Tech Department of Electrical and Computer Engineering
Images on this site are the property of Virginia Tech.
They may not be used for commercial purposes.
Last updated: Wed, Jun 25, 2003