Cover Page



Locally Networked Satellite-Based Computer Labs for Tanzanian Classrooms

Final Report

February 20, 2009

Sponsored By:

[pic]

In Cooperation With:

[pic]

University of Dar es Salaam Michigan State University

ECE 480 – Design Team #2 – Spring 2009

Management Brian Holt

Webmaster Daniel Newport

Document Prep. Steven Sadler

Presentation/ Lab Kevin Bishop

Executive Summary

With the increasing dependency on technology world wide there is a great demand to develop affordable personal computers for remote undeveloped areas. One region that may potentially benefit from this technology is rural East Africa, specifically Tanzania. The installation of computers and the integration of technology will allow the people of Tanzania a means of building a future for themselves, their families, their culture and their country. Before deploying a computer system into such harsh conditions several obstacles must be overcome including providing a reliable source of electricity to the system, telecommunications, and surviving the savannah climate. The Lenovo Corporation has tasked the spring 2009 team with the development of an integrated computing system equipped with Internet access that is able to accommodate up to eight users. The solution must be robust enough to withstand the harsh environment and must also be affordable for rural schools.

An uninterruptible power supply (UPS) will accept input power from the available power grid. Since the voltages often vary and the grid is subject to frequent outages the UPS system will be used to provide reliable uninterrupted power to the workstation. A point to point network will be created using a series of Patch and Yagi antennas to link Baraka Primary School to Manyara Secondary School. Satellite Internet will be shared between these two schools over the wireless link. The workstation will consist of two computers, and each computer will have the ability to support up to eight seats equipped with monitors, keyboards and mice. Each terminal allows independent simultaneous desktop sessions. The operating system we have chosen is Ubuntu Linux, and all other software on the system is open source.

This project is divisible into four equally important components that needed to be researched, tested, and built in order to make the project a success. The first section discusses power management as it relates to our system. Following this is a discussion regarding antenna design and specifications for creating a long range wireless link. The third section deals with advanced configuration of off the shelf wireless routers. The fourth section is a discussion of multi-seat system design and implementation, including Handin system specifications. System integration and project findings are addressed in the final section.

[pic]

FAST Diagram Outlining Functions of our System

[pic]

System Mock Up

Acknowledgement

We would like to first thank our sponsor, Lenovo, for allowing us to be a part of such a great project. We would also like to thank our team members in the Telecommunications department here at Michigan State University and the students and faculty that we have been communicating with at the University of Dar Es Salaam in Tanzania.

Also we would like to thank Gregg Mulder from the ECE Shop. His interest in antenna design provided us with a lot of useful information for prototyping and selecting the antenna that will allow us to connect the workstation to the internet successfully upon deployment to Tanzania.

This project would not have been possible without the help of our facilitator, Dr. Goodman. His dedication to this project provided us with motivation and kept us on track throughout the semester. We would also like to thank the members of Design Team 2 from last semester, their suggestions and advice provided us with a lot of insight to avoiding possible problems that may occur upon deployment to Tanzania.

Table of Contents

Section 1: Power Management…………………....Page 5

Section 2: Antenna Design & Configuration …..…Page 11

Section 3: Configuring the Router ……………......Page 16

Section 4: Computer Learning Environment……...Page 17

Section 5: System and Conclusion………………...Page 21

Appendix 1: Individual Roles/ Responsibilities…...Page 25

Appendix II: Literature and Web-site References....Page 28

Section 1 – Power Management

Determining the correct UPS type

There are three main types of uninterruptible power supply systems (UPS): Standby, Line Interactive, and Online. The Standby UPS is the most common type used for personal computers. In the block diagram illustrated in Figure 1,the transfer switch is set to choose the filtered AC input as the primary power source (solid line path), and switches to the battery/inverter as the backup source, in the event that the primary source fails. When this happens, the transfer switch must operate to switch the load over to the battery/inverter backup power source (dashed path). The inverter only starts when the power fails, hence the name "Standby”. High efficiency, small size, and low cost are the main benefits of the Standby design.

[pic]

Figure 1 – Block Schematic of a Standby UPS

The Line Interactive UPS, shown in Figure 2, is most commonly used for sensitive electronic equipment, critical servers and network devices. This type of UPS is able to tolerate continuous under-voltage, brownouts, and over-voltage surges without consuming the limited reserve battery power. When the input power fails the transfer switch opens and the power flows from the battery to the UPS output. With the inverter always on and connected to the output, this design yields reduced switching transients when compared with the Standby UPS. High efficiency, small size, low cost and high reliability coupled with the ability to correct low or high line voltage are the main benefits of the Line Interactive design.

[pic]

Figure 2 – Block Schematic of Line Interactive UPS

The Online UPS, sometimes referred to as a True UPS, is typically used for the same devices as the Line Interactive UPS. However, the Online UPS provides the highest quality power protection to electronic devices. This is done via a double power conversion technique. As can be seen in Figure 3, the Online UPS takes incoming power, converts it to DC, conditions it, and converts it back to AC. Since the input power is charging the backup battery source, which provides power to the output inverter, in the even of an input power failure on-line operation results in no transfer time. Excellent voltage protection and ease of paralleling are the main benefits of the Online UPS design. However, the double conversion introduces significant loss due to inefficiencies when converting from AC to DC and then back to AC.

[pic]

Figure 3 – Block Schematic of Online UPS

Summary of UPS Designs

The following table shows some of the characteristics of the UPS types described above. Since implementation and manufactured quality more strongly impact characteristics such as reliability, these must be considered in addition to the attributes highlighted in this table.

| |Practical Power Range |Voltage Conditioning |Cost per VA |Efficiency |Inverter Always|

| |(kVA) | | | |On |

|Standby |0 – 0.5 |Low |Low |Very High |No |

|Line Interactive |0.5 - 5 |Design Dependent |Medium |Very High |No |

|Online |5 - 5000 |High |Medium |Medium |Yes |

Figure 4: Table Comparing UPS Types. Note: Information in this table was obtained from, “The Different Types of UPS Systems” by Neil Rasmussen.

Determining Load Power Consumption

When determining which UPS system will best suit your needs it is important to first determine the amount of power the total system load will draw. Power is measure in watts:

P = V x I (Joule’s Law)

However, the power rating more often seen when selecting a UPS system is the Apparent Power Rating. Apparent Power is measured in Volt Amperes (VA):

S = V x I x Power Factor (Apparent Power Formula)

The power factor can be determined by taking the ratio of Power and the Apparent Power or as the trigonometric cosine of the angle between the voltage and the current. If the wattage is not listed specifically when choosing a UPS, a general rule of thumb is it is safe to assume that it is 60% of the Apparent Power. However, you should read the specifications of the UPS carefully because some systems will use a different ratio. The total load you attach to the UPS must never exceed either of its limits. It is recommended for most practical purposes that the total power in watts of your system is double the expected total load in watts. This ensures that there will room for expanded the number of devices connected, if necessary, and also ensures that there will be a significant amount of run time. Run time will be described in the next section of this document,

A list of all devices connected to the system should be made with their appropriate power draw, available from specification sheets provided with the devices themselves or from actual measurements. Figure 5 below contains such information for our system.

|Component |Power (Watts) |

|Lenovo S10 (Full Processor + Hard Drive) x 2 |(116.76 x 2) = |

|Satellite Router (Busty) |72.5 |

|17” LCD Screen x 8 |(20 * 8) = 160 |

|Linksys WRT-54GL Router |7 |

|Total |473.02 Watts |

Figure 5: Power Measurements on 11/11/2008 by Last Semesters Design Team 2. Note: Keyboards & Mice are currently not included in calculations because the team is not in possession of them and their power consumption is minimal.

Determining the Run Time

After determining the amount of power the total system load will draw, that number can then be used to determine the amount of run time the UPS can provide for the desired system. Run time is the amount of time that the battery backup will be able to power the total system load in the event of an input power failure. This number is often given in a table or chart, similar to the one shown below in Figure 5. The attached run time chart is for the APC Smart UPS 2200VA, the UPS we have chosen to power the workstation we will be installing in Tanzania.

|Watts |100 |

|Pin 1 – CS (Chip Select) |GPIO 7 |

|Pin 2 – DI (Data In) |GPIO 2 |

|Pin 3 – VSS (Ground) |GND |

|Pin 4 – VDD (3.3 Volts) |3.3 Volts |

|Pin 5 – CLK (Clock) |GPIO 3 |

|Pin 6 – VSS2 (Ground) |GND |

|Pin7 – DO (Data Out) |GPIO 4 |

Figure 8: Pinout for SD Card and Corresponding GPIO Ports on Router

Software

The router software for our design is OpenWRT. OpenWRT is an open source third-party firmware for routers. Using OpenWRT, we set up the router’s wireless as a Wireless Distribution System link (WDS). WDS allows two routers to be connected wirelessly; WDS is how we are providing internet to the second school as there is only one satellite uplink for internet. Also on the router, we attempted to use an MMC driver to communicate with an SD card to add memory to the router, but we were unable to get it working. There is no official MMC driver, all current MMC drivers are user developed with limited support. Some reasons for the SD card mod not working is that we are using are new release of OpenWRT and the MMC drivers may not be compatible yet. Another reason could be bad wiring or a bad card reader.

Section 4: Computer Learning Environment

Rationale

The One Laptop Per Child (OLPC) program aims to provide children in developing countries a laptop for their use. However, the major flaw with this program is that it fails to provide the infrastructure necessary for a cohesive learning environment. Lenovo Corporation, in partnership with Michigan State University, aims to solve this infrastructure issue by providing a holistic learning system capable of deployment in some of the world's most rugged areas. As a continuation of the efforts of the Fall 2008 design team we have elected to again use Open Source software to configure a centralized server into a multi-seat computing system. The benefit of this approach is that a centralized location can be used to teach the students, and thus teachers will likely be available to help guide them in a collaborative learning environment. Open Source software helps to keep the cost of the system as low as possible which is one of the primary design criteria. Three possible configurations (see the table below) were decided upon, and the Spring 2009 design team has chosen to use the third approach. Although the cost per seat is higher, by using only four seats on each PC, room is left for expanding the system at a later time. Two four-seat computers will be used in the system to be installed in Manyara Secondary school. Both systems will be mirrored using a piece of Open Source software called Unison which synchronizes file systems. Additionally, two more seats will be added to the system installed in Baraka Primary school by the Fall 2008 team bringing it up to a total of six seats. A great deal of testing was done on both the hardware and software to be used in the multi-seat system to ensure proper functionality.

|Criteria |Thin Client Based System |Multi-seat System with 8 Monitors |Two Server Multi-Seat with 4 Monitors |

| | | |Each |

|Low Cost |Lenovo ThinkStation S10 Server: $1300 |Lenovo ThinkStation S10 Server with |Lenovo ThinkStation S10 Server with 4 |

|(Priority: 4) | |4 Graphics Cards: $1300 |Graphics Cards: $1300 |

| |Lenovo L197 LCD | | |

| |Monitor: $239 each |Lenovo L197 LCD |Lenovo L197 LCD |

| | |Monitor: $239 each |Monitor: $239 each |

| |Keyboard/Mouse: $30 each | | |

| | |Keyboard/Mouse: $30 each |Keyboard/Mouse: $30 each |

| |Diskless Workstation Thin Client: $285 | | |

| |each |Cost per Seat: ~$431 |Cost per Seat: ~$594 |

| | | | |

| |Cost per Seat: ~$704 | | |

|Expandability/Flexibil|System is quite expandable, but at a |Possible to expand this system to |Use half as many seats on each server, |

|ity |higher cost. May need to upgrade the |contain up to 8 seats. Could use a |allows system to be expanded in the |

|(Priority: 2-3) |central server (costly) if too many |more powerful server to get more |future. Slightly more costly. |

| |clients are connected. |seats than this, still lowest | |

| | |overall cost. | |

|Low Maintenance |Performing maintenance may be difficult|Centralized point of configuration, |Two points of configuration; however, |

|(Stability) |on the thin client. |however, also a single point of |they are mirrored so complexity stays |

|(Priority: 3-4) | |failure. If system goes down there |the same. Even if one PC goes down |

| | |will be no computers. |there will still be 4 (or possibly |

| | | |more) seats available for |

|The user friendliness of each system is equivalent since almost any operating system can be chosen. Thus, the above three areas are the |

|main design criteria. |

Hardware choices and considerations

The hardware used in the multi-seat system is a Lenovo S10 ThinkStation PC which costs roughly $1300 with all of the necessary video devices for up to four seats. This provides a powerful backbone for running a multi-terminal computer. Any hardware chosen should be tailored to the maximum number of seats expected to be run off of the server so that over-engineering is avoided and costs can be kept at a minimum. One graphics card is needed for every two seats, and it is highly recommended to use Nvidia based GPUs due to their more Linux friendly drivers. Also, any graphics devices chosen should use the same driver software to avoid issues that arise from using multiple drivers (I.e. using two different Nvidia drivers). Additionally, only USB based input devices (keyboards and mice) should be used in the system. Through testing it was discovered that mixing input devices (PS/2 and USB) causes undesirable behavior. This issue will be discussed in greater detail below. The bulk of the work for configuring a multi-seat system is done in software, and is detailed in the following sections.

Multi-seat software setup and considerations

Two pieces of software are critical for successful configuration of a multi-seat system. The first is the Operating System, and the second is a piece of software called the Multi-seat Display Manager (MDM). MDM is optimized for use on Debian based Linux distributions, and we have opted to use Ubuntu Linux which is based on the Debian Operating System. Ubuntu was chosen by the Fall 2008 design team due to its user friendly interface, ease of use, and scalability. These features are important since many of the intended users have likely never seen or used a computer before, and it is highly desirable to keep costs as low as possible. MDM itself is not an actual display manager, but is used as a wrapper to create a multi-seat system based on the actual display manager. In this case MDM is used as a wrapper around the Gnome Display Manager (GDM). At boot time MDM configures the system by prompting users to press a Function key (e.g. F1, F2, etc.) followed by a left mouse click. This procedure binds that particular keyboard and mouse pair to the screen. The MDM method of implementing a multi-seat system works very well up to four seats. After more than four seats have been added some minor issues can arise. As noted previously mixing PS/2 and USB input devices can cause less than desirable behavior. MDM makes use of the Hardware Abstraction Layer (HAL) to determine which connected devices are input devices. More specifically HAL can be used to determine which connected input devices are keyboards and which are mice. The previously described issue results when more than four seats worth of keyboards and mice are connected. At this point HAL seems to have trouble recognizing PS/2 devices, and this can cause MDM to freeze unexpectedly during the multi-seat initialization. Thus, beyond four seats MDM only works partially. A fairly simple workaround to this problem is to limit the input devices to USB only, however, it is highly desirable for the multi-seat system to be hardware independent. Additionally, statically binding a keyboard and mouse pair to a particular screen without having to perform an initialization sequence would be a highly desirable future improvement and would make the system even more user friendly. Unfortunately this would not be an easy task since USB physical addresses can change based upon the order that devices are initialized at boot time.

Performance testing of a six headed multi-seat system was done by creating six user accounts and performing various tasks that could be expected during normal use. Normal use for the purposes of testing is defined to include word processing, browsing the internet, playing educational games, and viewing videos.

Handin system design

The intention of this system is that it be used in schools, thus it would be desirable to include a method for students to be able to submit assignments electronically to their teachers. Early in the semester it was decided that a Handin system should be built to accomplish this. Along with the ability to submit assignments this system should allow students to be grouped by classes, and have assignments corresponding to a particular class. Above all this system should be as user friendly as possible. The first implementation that was explored was Java based, but was not used because it would be too difficult to keep data persistent across multiple PCs. A program called ATACH (covered under the GPL) was found on Source Forge () which is a web based implementation using the Apache HTTP Server, PHP, and a MySQL Database. Although an excellent starting point, extensive modification of the original ATACH code, and restructuring of the database was required to make it usable for our system. One of the major modifications was a method for authenticating all users, and applying different classifications of users. Additionally, the original ATACH did not have the ability to submit files, thus this feature was added as well. The final system includes administrator, teacher, and student classifications. Students are assigned to a class by either an administrator or teacher, and homeworks are assigned at the class level. Templates can be uploaded for the students to use to complete assignments, and when complete students can submit their homework to the system for grading. Testing of the Handin system involved performing normal functions that the users would be performing. The most crucial test was submitting a large file to the database, downloading the file, re-uploading the file to the database, and handing the file back to a student. All of this was completed without noticeable loss in performance.

One of the most important features of the Handin system is that it is used as a way to integrate the entire multi-seat PC. When users are added to the Handin system they are also given an account on the PC that shares the same user name and password as their Handin account. All necessary files for use of the Handin system are created automatically so that no additional configuration is needed. This includes a Handin directory, an icon to open up the Handin system, and an appropriate user manual based upon the classification of the particular user. When users are classified as a student/teacher/administrator their system permissions are changed on the fly. Additionally, private-key SSH authentication can be used to create users on other PCs so that data will remain persistent. By using a web server the system can be accessed by anyone on the network, and thus a more sophisticated system could eventually be implemented in the future to encompass multiple neighboring schools. An advantage of this is that records could be maintained for each student as the progress through the school system to help better tailor to their learning needs.

Section 5: System and Conclusion

Integration

The final architecture that we chose was two workstations each equipped with multi-seat capabilities. This allows room for expansion of up to eight seats per workstation and will ensure that students will have access to a computer in the event that one of the workstations “crash”. We were successful in having up to 4 independent simultaneous logins from a single desktop computer with multiple graphics cards. We have made progress towards 8 independent logins per computer, as our current hardware can easily support up to 8 seats.

We were successful in combining our objectives into a complete and functioning prototype that we will deploy to a rural village in Tanzania. Our finished prototype has 4 user terminals connected to a custom designed enclosure that will contain two workstations and UPS.

Safety

Since we will be deploying this system to an underdeveloped area in Tanzania that is not familiar with computers and have little or no experience with electrical equipment we need to take extra precaution to ensure that the system is as safe as possible. There will be electrical connections traveling long distances inside the school so we must ensure that the cabling is run in some form of ductwork. This will prevent the cabling from being accidentally spliced and it will also keep all of the cables out of the way so that no one trips over them. The cabling will run from an enclosure that will contain the two CPU’s and the UPS to the eight monitors placed within the school. Placing the cabling in ductwork will also ensure that the cables are not accidentally disconnected from the equipment in the enclosure causing short circuits or possibly damaging hardware.

The UPS system also contains a large battery that if not properly cared for could cause a potential risk. We will secure the UPS system within the enclosure to ensure that they will not move or tip in the event that the enclosure is moved.

Having the UPS and computers enclosed will also eliminate the risk posed by having someone who is not familiar with the system pressing buttons and tampering with the hardware. However placing these components, which will have a lot of wires running between them, also poses a risk of short-circuiting the equipment. We must be very careful to keep wires leaving the case, and also within the case, organized to avoid the possibility of shorting wires. Any short circuit in the system has the potential to destroy expensive equipment or could possibly start a fire.

Since we will also be installing antennas on top of the roof of the school we must ensure that the insulation on the cabling can withstand various outdoor conditions including rain and extreme heat. If these cables are not properly selected the insulation could wear, leaving bare wire exposed.

Designing this system with these precautions in mind should keep the system safe from being, weather, and equipment failure. If troubleshooting is necessary, we will also have remote access to the system that we install via software installed on the computer system and also through logs created by the UPS system. This will allow us to monitor the system and make it easier to diagnose any problems that may occur once we have installed the system. For instance if equipment has been disconnected from the UPS, the software that creates logs for the UPS will show a difference in the total load wattage of the system and allow us to easily determine that this is the problem.

Future Improvements

A major improvement that could be made to the system is choosing a computer system that is optimized for minimal power consumption yet is also within the requirements for the Multi-Seat setup. The computer we are using in the prototype is the Lenovo S10 workstation. This computer is designed to be a state-of-the-art processing power house, and was not necessarily designed for efficiency. However, this is the only hardware that Lenovo currently offers that would support the Multi-Seat architecture; therefore it was chosen for use in this proof of concept prototype.

Another area in which improvements could be made is the number of seats available in the Multi-Seat workstation. We currently have the workstation setup with seats available for 8 students, 4 from each workstation. We have restricted the number of seats available to 8 mostly because we are certain that we have 8 monitors available to us in Tanzania. This number was also chosen to maintain a runtime of at least one hour. Once we are onsite in Tanzania we may expand the system to allow more users. However, this will not happen until we have successfully set up the currently planned system.

|Items |Cost |Quantity |Total Cost |

|Lenovo Nvdia Quadro |$145.53 |1 |$145.53 |

|GeForce 8400 GS Video Card |$63.99 |2 |$127.98 |

|Linksys WRT-54GL Router |$64.98 |4 |$259.92 |

|N-Female to RP-TNC Male Plug Adapter |$4.98 |6 |$29.98 |

|15ft. SVGA Cables |$6.99 |8 |$55.92 |

|Kingston 2Gb SD Memory Card |$7.98 |1 |$7.98 |

|2.4GHz Wifi Yagi Antenna |$29.95 |2 |$59.9 |

|Connex Wireless Q-Bridge+ Antenna |$399.99 |1 |$399.99 |

|Lenovo S10 Workstation |~$1300.00 |1 |~$1300.00 |

|Miscellaneous Cables/ Plug Adapters |$138.30 |---- |$138.30 |

|APC Smart-UPS SUA2200I** |$1,094.40 |1 |$1,094.40 |

|Suntech 60W Crystalline Solar Module** |$422.00 |1 |$422.00 |

|Morningstar 6a Solar Charge Controller** |$70.00 |1 |$70.00 |

|Containment Box made to specific design** |$450.00 |1 |$450.00 |

| | |Total |$4,561.90 |

Figure 9– Team Budget. Note: Budget is Based on Items That Will Be Part of the Deployed System; Items Denoted With ** Were Purchased in Tanzania

Conclusion

The Lenovo project is unique because it is not entirely focused on the Design Day presentation. A more important aspect of this project is the successful deployment of a system that will provide an educational resource to Tanzanian children who are currently living without access to these technologies. Quality education is a pillar of national development and the Internet is an extremely valuable asset that people in Third World countries often do not have access to. Through quality education Tanzanian citizens will be able to create a means to effectively cope with the challenges of development and confidently adapt to the changing market and technological conditions not only within the region but also globally.

This May, Michigan State University’s Design Team 2 will deliver the system it has designed to Tanzania. The installation of this system will have a great impact upon the lives of the school children.

Appendix 1: Individual Roles/ Responsibilities

Daniel Newport:

My role in team 2 was the configuration, testing, and implementation of the multi-seat system. As a result of this project I have gained valuable experience both professionally and technically. Most of the work I have done this semester was split between the computer system configuration, and writing a Handin utility for use in schools using the system. The conception of the Handin system is that it be used as a way to integrate the entire computer in an effort to make it more user friendly. Although a significant amount of effort went into the coding and the testing of this system, there are still some minor bugs and fixes that could be made to make it even better. I am hoping to have time to revise and optimize the system over the summer to make it even better.

Steven Sadler:

I was this semester’s document preparer. I handled the formatting and finalization of all documents submitted by the team. My primary responsibility was researching and testing of possible solutions to the lack of reliable power in the Tanzanian school. My research began with intentions of building our own power source consisting of an appropriately sized battery bank connected to an inverter/ charger combination. However, after researching this idea I realized it would not be feasible due to the problems with extreme varying input voltages we were expecting to see and the protection required for the computer system. This lead to researching and choosing an Uninterruptible Power Supply (UPS) capable of handling varying input voltages, a 500 Watt system load, and that was available locally in Tanzania.

While it was not my primary responsibility during the course of this project, I also helped the team by researching antenna types and capabilities of wireless routers. Brian and I built and tested a prototype Quagi antenna, which is a combination of a Yagi and Quad antenna designs, and obtained encouraging results that led to the purchase of two Yagi antennas that will be used as part of the network link in Tanzania.

Brian Holt:

My technical responsibility has been to design, choose and test antennas for use in transmitting and receiving wifi signals. This entailed, first of all, research into the feasibility of sending a wifi signal of 2.4GHz over a distance of approximately two miles. Concluding that it is, indeed, possible to send a 2.4GHz signal such a distance, the next step was research antenna theory and design. This turned out to be a fascinating topic. There are a variety factors that must be taken into consideration, such as gain, target frequency, polarity, power, radiation pattern, etc. Antenna design itself, however, is relatively simply.

Having become familiar with the basics of antenna design, I proceeded to the selection process. Among the types of antennas considered were yagi, quad, quagi, helical, antenna. Conferring with other team members, we decided to build a quagi antenna. I constructed the first prototype antenna using easily accessible materials, such as copper wire, a wooden road, coaxial cable and a connector. Detailed information on how this was constructed is provided above. After testing the antenna successfully at short range, Steve Sadler and I built a second quagi antenna following the same design.

Driving off campus, we tested the designed antennas on three occasions. The first time the weather was bad (cold and snowing). The second time, as well the first, it turned out that the router was pre-programmed only for short distances. The third time, with the routers reprogrammed, the antennas did quite well, all considered; however, the antennas we had purchased outperformed the designed antennas.

After the first two attempts to achieve acceptable distances, I selected, and we purchased, two15 dBi yagi antennas (manufactured by MFJ) for testing. These antennas were priced more reasonably ($29), and had a higher dBi, than other comparable alternatives. Driving once again off campus, we performed two additional field tests. We were able to achieve a signal range of approximately 1 mile.

Surprised at the results, given what my information had led me to expect, I conducted further research on ways to extend the distance. Two options for increasing antenna range are to increase power input (power amplification) and to increase gain (choosing a different type of antenna). Neither of these options, however, as it turns out, are acceptable, given the legal restriction of 4 watts effective isotropic radiation power. The only other option, then, was to find another type of antenna, with the same input power and gain, capable of sending an acceptable signal at least 2 miles. Accordingly, I found, and ordered, a set of antennas meeting the legal requirements which, it is claimed, can send a signal up to 4 miles. These are Q-Bridge patch antennas with 15dBi gain made by connexwireless, designed specifically to send and receive wifi signals over long distances.

Kevin Bishop:

My technical role on the team was to research, program and set up the routers. I researched many routers to find one that could be reprogrammed with third-party firmware, was cheap, and had removable antennas before finally selecting the Linksys WRT54GL. Programming and setting up the routers included installing OpenWRT on all the routers, configuring wireless and DHCP settings, troubleshooting routing issues, doing a hardware modification on one router to install an SD card reader, and installing software to use the card reader. There are 4 routers total; I flashed each router with OpenWRT release 8.09. I researched how to connect routers together wirelessly, and found that Wireless Distribution System capabilities were needed. Since OpenWRT supports WDS, I configured the two routers that will be using antennas to use a Wireless Distribution System link. I disabled DNS and DHCP on all but one router to avoid addressing problems. The DHCP, I set to assign IP addresses in the range 192.168.1.50 to 192.168.1.100 to limit the number of computers and to leave lower addresses open to be statically assigned to routers. For each router, I statically assigned one of the lower IP address not in the DHCP range to avoid getting duplicate IP addresses on the network and to ensure that the router IP addresses would not change. To install the SD card reader, I took apart a USB SD card reader and removed all of the main components except for the card slot. I then soldered each pin of the card slot to the router mainboard. I researched the driver needed to access the card and installed it on the router. I installed several different versions of the driver to try and get it to work, but was unsuccessful to this point. I also worked on setting up the Q-bridge antennas so they worked with our routers and received signals between each other.

Appendix II: Literature and Web-site References

•

•

•

•

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download