Xavier Halloran


Computing Research Topic Selection


Topic:

Wireless Distribution Service for Large Scale - Low Power Networks.


Description:

In many situations, the deployment of computer networks is critical, yet establishing a wired network is unfeasible.  In this case, completely wireless networks, or networks with no significant distance wiring, are required.  The arrangement and configuration of the many access points and the possible protocols are areas of interest.  Directed mid-range wireless signals, on the order of hundreds of feet, create long-range noise, polluting valuable and limited spectrum.  Sharing a wireless spectrum between the back-bone signals, what would be the equivalent of the wired part of a typical wireless network, and the users becomes yet another problem.  Data routing in this case is simplified, as all of the access points are geographically static, and paths are constant.  Power supply is an issue, as the access points may be placed in areas where no steady electrical supply is available.

This paper does not address a network protocol.  Since this network is simply a wireless version of a wired network, with static access points, data control and routing is already extensively researched.  Utilizing already known routing information, such as developed by Cisco routers, this network is simply an extension of existing information.



Motivation:

Many areas require reliable internet connectivity spread over large geographic areas, in regions where typical wiring is unreasonable.  Disaster relief and coordination, as well as general long-term support for slow-recovery areas, such as Chalmette and St. Tammany, Louisiana, where there is still no or very limited electrical power and wired communication, are possible applications.



More Information:

The technology for Wireless Distribution Service already exists, though public adoption is still relatively rare.  WDS creates a wireless backbone between access points in the same wireless network.  It works as though it were a traditionally wired network.  New IEEE specifications allow for static routing of data though such a network.

WDS devices act in two ways, either as a standard access point for wireless devices, a bridge connecting access points together, or both.  DD-WRT firmware for the WRT-54GL allows WDS, and supports mapping.  This allows a cyclic loop of WDS stations without a data cascade, that is, data will not infinitely travel though the network.  Looped networks offer fault tolerance, a dropped link will not cause a network failure.

WDS halves network bandwidth per unit, as each signal needs to be received then retransmitted.  Clever channel distribution and multiple gateways can reduce the total amount of network traffic.

DD-WRT firmware allows adjustment of broadcast strength, allowing a reduction in power consumption while still maintaining a clear signal.

Extensive research on antenna optimization exists, though are not applicable to this project.  The purpose of this project is to keep costs as low as possible, and homemade antennas are not as finely-tuned as professional equipment.  The exact measurement required for perfect transmission is not achievable on homemade equipment.


Problems:

Academic papers describing battery power supplies are limited.  There are many discussing theoretical power and power supplies, along with ways to generate power.  Using batteries, solar panels, and AC power to power the devices is not a matter of academic research, rather a matter of applied electrical engineering.

Directional antennas, particularly D.I.Y. directional antennas, have been known to reach 10 mile links.  This is extremely useful to create large-scale networks, however, testing and development is complicated, isolating the units to test are hard to do with signal range exceeding several hundred feet.  Also, links shorter than this, which will be the general case, will cause noise at great distances and cross talk.  It does also allow for redundancy, a node failure will not be a problem if the devices can simply talk over it.


Possible Implementations:

At each physical location, two network devices can be used.  Each WRT-54GL has two antennas, and with two devices a total of 4 antennas can be delivered to each location.  One device can act as a WDS, using a specific channel, connecting the two neighboring nodes, and the other local WRT device.  The WDS device can have two highly directed antennas, linking up and down the chain to other WDSs.  The other local device, acting as an AP, is fitted with two biquad antennas.  This makes it a focused antenna, nearly 50 degrees.  Clients attempting to access the network can connect to these devices, and the overall network coverage can be increased if the clients also enhance their signals with directional antennas.  Network bandwidth can be increased if the AP and WDS act on different channels.  The AP and WDS devices can be wired, their distance apart is negligible, about 2 inches.




References:


M. Burkhart

"Analysis of Interference in Ad-Hoc Networks"

Diploma Thesis

2003

This article addresses issues of signal interference in networks, as the title suggests.  Addressing multiple angles, such as interference at transmission or at reception, it seeks to minimize excess radio signals.  It does not, however, address the possibility of directed radio signals, yet this does not diminish its usefulness.  Individual nodes in the network are undirected, and still need to be addressed.

The article offers an interesting example of a campsite border each side of a river.  It analyzes the possible radio setups and potential ways to minimize signals.



Building Robust Wireless Mesh Networks Using Directional Antennas: How Many Radios Are Enough.

Qunfeng Dong, Yigal Bejerano, Suman Banerjee 

http://www.cs.wisc.edu/~qunfeng/papers/UW-CS-TR1572.pdf

This article addresses directional antennas in mobile applications in order to improve signal range.. By restricting the signal transmission, it consequentially reduces the signal reception as well.  In the arrangement proposed by this paper, directional antennas provide great power, narrowing the noise in single directions, and preventing signal noise.  However, because of the narrow range of transmission, multiple antennas will be needed at a single point, creating a spoked effect.

Furthermore, this article addresses the concept of multiple gateways.  With more than one way onto the larger internet, network traffic can be reduced.  Total network distance can be cut in half with well placed access points.



Turning a $60 router into a $600 router

Aaron Weiss

November 8, 2005

http://www.wi-fiplanet.com/tutorials/article.php/3562391

This gives directions on how to modify the firmware on a WRT 54 GL Linksys router.  The router is one of the most flexible wireless devices on the market, running open source Linux.  There are multiple versions of firmware with multiple applications.

"You could use the WRT54G as a repeater or a bridge. Create a wireless distribution system (WDS) or a mesh network. Run a VPN server. Or a VoIP server. Or a managed hotspot with a RADIUS server. Manage bandwidth use per protocol. Control traffic shaping. Support IPv6. Boost antenna power. Remotely access router logs. Operate the router as a miniature low-power PC, running a variety of Linux applications."


LinksysInfo.org, WRT54G.net, (and strangely enough, en.wikipedia.org/wiki/WRT54G)

All address the issues of installing, operation, and manipulation of WRT-54G routers.  By manipulating the firmware and establishing a customized operating system, the devices can be controlled in any number of customizable ways.  Establishing sleep cycles, selectively utilizing antennas, and adjusting transmission power are all possible via the custom firmware, and can even be controlled remotely via a web interface.



Delay Efficient Sleep Scheduling in Wireless Sensor Networks

Gang Lu, Narayanan Sadagopan, Bhuskae Krishnamachari, Ashish Goel

http://ceng.usc.edu/~bkrishna/research/papers/LuSadagopanKrishnamachariGoel_Infocom05.pdf

This paper addresses a mechanism for communication in wireless sensor networks adapted for low power consumption.    It can easily be adapted to a wireless tree network, and such a network is in fact addressed as a special case.  The caveat with this article is the base assumption of low traffic in order to address the issue, while this research proposal assumes that the network may potentially be crowded with traffic at certain times.



"Capacity of Ad Hoc wireless networks" International Conference on Mobile Computing and Networking, Proceedings of the 7th annual international conference on Mobile computing and networking

Pages: 61 - 69

Year of Publication: 2001

ISBN:1-58113-422-3 

Jinyang Li, Charles Blake, Douglas S.J. DeCouto, Hu Umm Lee, Robert Morris




On the capacity improvement of ad hoc wireless networks using directional antennas

International Symposium on Mobile Ad Hoc Networking & Computing

Proceedings of the 4th ACM international symposium on Mobile ad hoc networking & computing

Pages: 108 - 116  

Year of Publication: 2003

ISBN:1-58113-684-6

Su Ui, Yong Pei, Shivkumar Kalyanaraman

The article discusses the ability to improve wireless network traffic by directing the signals in only the required direction.  Omnidirectional antennas radiate a signal in all directions, and can cause interference in areas where the signal noise could otherwise be eliminated.  Overlapped nodes can still comunicate with other nodes without intereference if the directional antennas are not aimed at each other.  For example, a North-South oriented antenna will not lose efficiency if an East-West signal should overlap it.  However, because of this, the E-W device will not be able to directly talk to this specific N-S oriented device. 




Wireless Networking Info, Biquad Antenna Construction

http://martybugs.net/wireless/biquad/

Martin Pot (mpot@martybugs.net)

22 November 2006

This site offers detailed design plans for constructing a bi-quad antenna optimized for WiFi spectrum.  It shows step by step construction and all parts needed in order to build one and compares the results to commercially produced antennas.


http://www.trevormarshall.com/biquad.htm

Trevor Marshall

BiQuad 802.11b 11dBi wide band antenna

Offers similar homemade antennas for use with WiFi, made with parabolic reflectors to direct the signal in a more refined manner.  Also shows performance gain information and comparisons to commercial equipment as well.



"An energy efficient MAC protocol for wireless LANs"

Eun-Sun Jung   Vaidya, N.H.  

Dept. of Comput. Sci., Texas A&M Univ., College Station, TX;

This paper appears in: INFOCOM 2002. Twenty-First Annual Joint Conference of the IEEE Computer and Communications Societies. Proceedings. IEEE

Publication Date: 2002

Volume: 3,  On page(s): 1756- 1764 vol.3

ISSN: 0743-166X

ISBN: 0-7803-7476-2

INSPEC Accession Number: 7492187

Digital Object Identifier: 10.1109/INFCOM.2002.1019429

Posted online: 2002-11-07 17:05:04.0 



"Wireless sensor networks for habitat monitoring"

nternational Workshop on Wireless Sensor Networks and Applications;  Proceedings of the 1st ACM international workshop on Wireless sensor networks and applications

Atlanta, Georgia, USA

SESSION: Applications and OS table of contents

Pages: 88 - 97  

Year of Publication: 2002

ISBN:1-58113-589-0 



"PEAS: a robust energy conserving protocol for long-lived sensor networks"

Fan Ye   Zhong, G.   Songwu Lu   Lixia Zhang  

Dept. of Comput. Sci., California Univ., Los Angeles, CA, USA;

This paper appears in: Network Protocols, 2002. Proceedings. 10th IEEE International Conference on

Publication Date: 12-15 Nov. 2002

On page(s): 200- 201

ISSN: 1092-1648

ISBN: 0-7695-1856-7

The published versions of this paper are only two pages, and offer only cursory information.  A 12 page version exists, but it has not been published to any public archive.

The paper discusses a low energy wireless protocol for sensor networks.  It utilizes a large number of small sensors with short battery lives.  It saves power by cycling off redundant nodes and maintaining only what is required to sustain a signal.  It does not need to know the geography of the nodes and observes only its immediate environment, it will not keep track of the states of it neighbors.  According to the article, it can sustain adequate density even if up to 38% of the nodes fail.  It can maintain a nearly constant overhead level and uses only about 1% of total energy use.

As a caveat, the network topography proposed by this research proposal utilizes a sparse network with a small number of sensors, the exact opposite of the assumption of the article.