Wireless Roaming Access Point
October 16, 2007
Faculty Technical Advisor:
Prof. Hong Man
I pledge my honor that I have abided by the Stevens Honor System
Table of Contents
II. Project Proposal Plan
2. Design Requirements…………………………………………………………………..4
3. Design Approaches…………………………………………………………………..4-7
4. Financial Budget……………………………………………………………………8-10
5. Project Schedule…………………………………………………………………..11-12
Many WLANs are widely based on IEEE 802.11 standards. Research has been
relentlessly conducted in order to improve the quality of signal strength. Even in high-
speed systems such as IEEE 802.11a/b/g, traffic congestion is still a major barrier that
degrades consistent connectivity for hosts attached to the access point (AP). Most
networks that have this problem are large and employ multiple APs. Another solution is
to have the APs intelligently to determine the size of the network. Under these congested
conditions, intelligent APs are required to communicate with each other and determine
the number of hosts connected with their respective bandwidth usage. A select number
of APs will have their firmware modified so that the coverage range will be
autonomously changed to balance the overall traffic. Simulations will be conducted in
order to test the firmware once it is ready. Performance results will also be recorded and
presented in the final report.
II. Project Proposal Plan
Answers to improved signal strength, thus far, have been focused more toward a
single access point as opposed to the collaboration of a number of routers. In the
research paper titled "Intelligent Radio Resource Management for IEEE 802.11 WLAN,"
by J. Bigham et al, the solution to improved overall signal strength was a four-sector
antenna array that fluctuated power if data rate was deemed "traffic overload" when the
rate was over a certain threshold. The group's project will address the problem quite
differently in that the data rate will be communicated between separate APs. This will
allow the APs to distribute their loads amongst each other.
The Linksys WRT54G series provided
by Cisco is a Wi-Fi capable residential gateway.
With 4 ports for 802.3 Ethernet and IEEE
802.11b/g capabilities, the router provides
sharable internet connections using the same IP
address. The WRT54G router is considerably
noted for its firmware source code released so that users may change or add utility to the
The DD-WRT open source firmware, used by this type of WRT54G series, is
presented in user friendly GUI format and the control panel presents many options for
customization. The following picture shows a sample of the many properties capable of
MAC addresses identifying the different users and their respective signal-to-noise
ratio are also conveniently displayed. This information will be crucial for AP
communication to implement range fluctuations. The WRT54GL router uses a Linux-
based Operating System as opposed to the previous versions that used the VxWorks
Because the provided firmware allows programmers to customize many
properties, it requires a lot of time to explore into the many directories pertaining to each
II-2 Design Requirements
There are many constraints to be taken into consideration while designing a
network of wireless access points (WAPs) to dynamically transfer hosts between one
another. For example, one must take into consideration the power limitations of WAPs,
understanding that there is a lower bound at which a host will receive a signal; one must
also realize the difficulties associated with passing a traffic-laden host between access
points – while maintaining a pipeline through which the user can keep a seamless
connection. To this end it is necessary to keep track of hosts that have already provided
authentication credentials and “logged in,” such that users do not need to re-authenticate
(this would create a gap in the end-user’s connectivity, perhaps costing the user
invaluable resources). It is also quite necessary to have a system by which an access
point can determine which hosts are utilizing vast amounts of traffic - and where they are
located; this should also be maintained in a table of sorts. Finally, to facilitate all of this,
it is useful- if not wholly necessary- to acquire a device with a firmware environment
within which one can easily interface and write code. Without this interface, the project
faces the risk of failure.
II-3 Design Approaches
There are numerous design elements needed to achieve the stated goal of the
project. The elements fall under two very general categories: ones that will be constant
throughout all the design approaches and ones that the group could change to achieve the
outlined design requirements.
The concrete elements are as follows:
1. The project will consist purely of augmenting existing code in open-source
firmware called DD-WRT, which will be installed on Linksys WRT54GL
routers. This approach will ensure that the network will be 802.11g
compatible and require no software installation for the end-user.
2. The code will control the power usage(transmitter power); therefore, the
effective broadcast radius of three routers in a cell network.
3. Each router will broadcast on a different channel available within the 802.11g
protocol automatically so that overlapping cells do not interfere with each
Specifically, the group will use the generic Windows descriptions for signal
strength in its algorithm for determining which nodes should be handed off. Both the
connected node and the access point measure strength in –dBm, so naturally the AP’s
value will be used. The descriptions break down as follows:
-dBm quality -dBm quality
<60 excellent 60 – 64 very good
65-69 good 70 – 79 low
>79 very low
During the testing for signal strength noise levels remained constant, and will not
be accounted for at this point; however, the noise levels are displayed along with the
signal strength readings, and the group will keep an eye on them to see if they cause a
problem as the system increases in scale (with more access points and user nodes).
802.11g accounts for noise somewhat anyway.
Another metric needed is the effective range of the router. The transmit power
usage of the router is the prime and easiest to control metric regarding the range. So far
the group’s tests have been inconclusive as the test space was not large enough to get
useful data to establish a specific trend between power and range. The maximum test
distance was approximately 25 ft and all test points up to that distance at all power usages
were greater than -60 dBm and in the excellent range. The plan to gather such data is
now to use the bowling alley after hours to allow the required space. The WRT54GL can
vary its power usage for transmission from 1 mW to 250 mW in integer increments.
Default power is 28 mW.
Four network topologies are being considered for the implementation of the
system. The ideas will be judged on the criteria: ease of design, ease of end-user use and
installation, cost, scalability, and effect on network throughput. Regardless of the
implementation, the network needs to store tables that consist of AP and node addresses,
network verification, percentage of maximum throughput, current signal strength, simple
signal strength history, approximate geographic location, and connection time. Using
these metrics the network decides which APs are overloaded and underloaded, and which
nodes require more throughput. Then the network tells APs how to adjust the size of
their range and distribute the nodes to maximize overall throughput and power efficiency.
The first involves wiring a dedicated server with cat5 that will perform the logic
and control for all access points. This topology allows for maximum throughput and
control, as no control messages will cut into the 54 Mbps Wi-Fi throughput, allowing
nodes full access to the Wi-Fi and as many control messages as wanted to be sent without
cutting into the Wi-Fi. The topology will be costlier due to the need for the server and
running the wired media.
The second is similar to the first except the dedicated server communicates with
the APs through the Wi-Fi. The dedicated server broadcasts only on the channel which
the destination AP is on. It will be easier to set up, but will cut into the Wi-Fi throughput.
The dedicated server will still minimize Wi-Fi greediness, because it will only
communicate to the AP that needs communication.
The third topology will not have one dedicated server, but will have one AP
(more in a scaled up system) set as a master and the rest will be slaves, taking direction
only from the nearest masters. In a larger network, the masters will communicate with
each other to coordinate their tables and avoid conflicting instructions going to APs near
The last topology shares the logic and control between all the APs. It is very
scalable and new APs could be added to the network easily. The overhead for the control
would be a limiting factor as an AP would need to poll all its nearest neighbors to know
which would best be suited for receiving a higher load. Or the APs would have to be
frequently sending updated table to each other to maintain full knowledge of the status of
the network. Either solution would require much use of the Wi-Fi capabilities cutting in
the 54Mbps data rate.
Roughly sensing the geographic location of the nodes will be necessary no matter
what, but the precision of the location will vary depending on how the APs are set up. If
the APs are set up close enough to the boundary of the desired coverage area, then a
minimum radius can be enforced so that only the radial distance needs to be measured
(through signal strength). If the APs are further away from the boundary then
triangulation will need to be used to detect the radius and the angle so that a node on the
opposite side of the overlap region will not lose connectivity by an AP reducing its range
to a point where the node is no longer covered.
II-4 Financial Budget
The budget for the project has been allocated as per the figure on the following
page. This initial configuration is based on a minimal setup which minimizes the project
costs based on implementations considered so far. If we are able to meet our design
goals with great forward momentum then we may need extra resources (e.g. sectored
antenna) to add additional features.
Description Quantity Price Total
Team member salary 0 $0 $0
Materials and Parts
Linksys WRT54G Wireless access point 3 $66 $198
Laptops 4 $0 $0
wireless adapters 4 $0 $0
Ethernet cable 4 $0 $0
DD-WRT / OpenWRT firmware $0
Linux based OS $0
printing paper 500 sheets $8 $8
ink 1 cartridge $20 $20
Commuter expenses $20 $20
Assistance from Faculty Advisor /
Professors $0 $0
Rent, Utilities, Overhead
Telephone costs $0 $0
Senior design funds $250
Sponsor funds/grants Undetermined
The budget represents the incurred costs to the group in executing a proof of our concept.
The budget does not consider the group members’ tuition and laboratory fees, which
minimizes the labor costs to zero. The major parts required for this project would the
Linksys WRT54G Wireless access points to setup our initial three node structure. The
laptops, wireless adapters, and Ethernet cables are available as the school issue standards.
The firmware and OS are open-source software so that the group doesn’t incur a charge
from their use and MATLAB is issued with the software package provided to students.
The budget allocates most of the $250 provided for senior design ($4 unallocated).
Sponsor and additional funding is undetermined at this point but the group hopes to
contact companies and universities interested in effective radio management to ask for
advice or funding in the near future.
II-5 Project Schedule
The group’s project schedule is represented by Gantt chart for senior design
(encompassing semesters seven and eight). Programming different aspects of power
management and software development is a major part of this project. The tasks are
designated to different team members as represented in the resource names column.
Everyone will have some part in each aspect while the dominating influence will be the
members leading the specific tasks as previously mentioned. The schedule is expected to
be changed as problems are expected and new tasks emerge.
The Wireless Roaming Access Point fluctuate its range according to individual
throughputs of each AP involved. This project will lead to the production of intelligent
APs capable of collaborating with others to ensure the best possible quality of internet
connection. Success of this project will result in higher throughput and better power
efficiency for the network of routers considered. In ideal situations, businesses requiring
many APs will benefit from the project considerably if the devices collectively
encompass the entire region. Due to the inexpensiveness of the project, businesses are
not required to utilize funds on expensive appliances suited to improve signal strength
such as repeaters. Users are also not required to continually manage the properties of the
AP manually as the device will establish best working conditions autonomously.
Ultimately, the successful device will increase throughput and substantially reduce end
Bultitude, Robert, Robert Davies, and Robert Hahn. "Propagation Considerations for the
Design." IEEE, 47.1(1998). 23 Sep 2007
"DD-Wrt." firmware features. 3 Oct 2007 <http://www.dd-
"Doppler Spectrum." Welcome to Wireless Communication. 23 Sep 2007
Friday, Bob, and Alex Hills. "Radio Resource Management in Wireless LANs." IEEE
Radio Communications, 10 (2004). 9 Oct 2007
"HyperWRT - Features." HyperWRT - Home Page. 1 Oct 2007
"Linksys.com - Products/Wireless/Basic Networking/Broadband Routers/Wireless-G
(802.11g)/WRT54GL." Welcome to Linksys.com. 30 Sep 2007
Moustafa, Habib, and M.N. Naghshineh. "Efficient radio resource control in wireless
networks." IEEE Transactions on Wireless Communications, 3 (2004). 7 Oct
Peh, Li-Shiuan, and Vassos Soteriou. "Dynamic Power Management for Power
Optimization of Interconnection." IEEE, 11 (2003): 6. 24 Sep 2007
Weiss, Aaron. "DD-WRT Tutorial 2: Extend Range with WDS." Wi-Fi Planet - The
Source for Wi-Fi Business and Technology. 5 Oct 2007 <http://www.wi-
Weiss, Aaron. "DD-WRT Tutorial 3: Building a Wireless Bridge." Wi-Fi Planet - The
Source for Wi-Fi Business and Technology. 16 Oct 2007 <http://www.wi-