Networking

University deploys a WLAN

Lessons learned in WLAN effort


When one of our on-campus business units had outgrown its building, my boss came to the IT department and said that the company had decided to lease another building down the road and move this business unit to the new location. He asked IT to look at the options for connecting the building to the corporate network and said that they would probably need a faster connection than they have now.

The unit's current building was located on campus and connected via a wireless Ethernet bridge. The link was installed four years ago and ran at 2 Mbps. Our first thought was that a faster wireless link would work for the upgrade, but we also knew that we would have to calculate and justify the cost.

To give you a better understanding of a wireless bridging deployment, I'll explain the process we went through to estimate, prepare, and deploy this link. I'll also compare the cost savings of this solution versus a wired installation.

Planning
We first contacted a local wireless vendor and discussed various options. The vendor told us about Proxim's Tsunami wireless bridges product line, which could carry data at speeds ranging from 10 Mbps to 1 Gbps. The product line also included a feature for adding a "wayside T1" for voice.

Initially, the project was only for data, but when we learned about the voice option, we decided to extend the current voice system via the wireless link as well. The next step would be to perform a site survey to check for obstructions and to get a rough idea what heights would be involved.

For a cost comparison, we also contacted the local telco provider and discussed various options using multiple T1 lines or a partial D3 circuit. We decided initially to compare costs on a 10-Mbps leased-line connection to a 10-Mbps wireless bridging solution.

Site survey
The two buildings were located roughly a mile apart. Between the two locations lay farmland and woods. With woods come trees and trees can be a nightmare for wireless signals. To make the link work, we would have to be above the trees (since wireless bridging requires an unobstructed line-of-sight between the two units). This would require a tower at both ends of the link.

Fortunately, we already had a 150-foot communications tower located near the main campus building. After climbing our tower and visually surveying the path, the vendor recommended a 100-foot tower at the remote end of the link.

We contacted the management of the new building, who said that a 100-foot tower was out of the question. They said we could mount our equipment on the building's roof as long as it wasn't visible from the road. While doing the site survey, we went with the wireless vendor up to the roof of the new building. We located a spot in the middle of the roof, but the vendor suggested a spot near the edge that had some open area. There was just one problem: From that spot, a single tree obstructed part of the path to the campus tower. Although a tree would normally be a problem, this tree appeared to be dead. The vendor explained that the tree leaves are usually the main transmission barrier because they contain water, so the dead tree may not cause an obstruction. The vendor representative said that they would be willing to try it but offered no guarantee.

Cost analysis
The wireless vendor submitted a quote for the installation. Figure A shows the costs compared to the phone vendor's quote.

Figure A


As you can see, the initial equipment costs and installation were about the same for both systems. The real cost savings in a wireless system are realized because there are little to no monthly recurring charges. The only recurring cost you may encounter with a wireless installation is tower rental. In this installation, we owned the hospital tower, so we had no monthly charges there, and our initial wireless quote included the cost of purchasing a 100-foot tower for the remote location.

In addition, since the 100-foot tower was out of the picture for the remote building (because we simply mounted the equipment on the building's roof), we decided to use the money we saved there to purchase 45-Mbps equipment instead of the 10-Mbps equipment we had originally quoted for the link.

Installation
The target date for the installation was December 31, but the project was delayed due to construction problems and was pushed back to February 14. This gave us more time to work but lousier weather to work in. Normally, our winters are pretty mild, but not this one. The day the antenna was installed on our tower, it was around 0 degrees. Despite the weather setbacks, the antenna and cable were mounted successfully on our main tower. The other end of the link was completed the next day.

After both ends were completed, the vendor performed some fine-tuning of the antenna alignment. The link was fired up and everything appeared to be working well. We were installed and running two weeks ahead of schedule.

Disaster strikes
My colleague and I monitored the link for the next several days. The Tsunami bridges have a Web-based management page that gives the status of the link and indicates any alarms (Figure B).

Figure B


During one of those first few days, I was sitting in my office watching a heavy snow out my window and decided to check the link. Much to my surprise, it was showing a high error rate and occasionally would drop completely. I immediately placed a call to the vendor.

He explained that rain or snow would not be a problem, but if the snow was sticking to the dead tree and forming a "wall of water," that could be an issue. He promised to be out Monday to try to determine firsthand what was going on. I watched the link off and on over the weekend. It was very sporadic and didn't seem to have any correlation with the weather.

When the vendor arrived on Monday, we insisted that the antenna at the remote location be moved to the original location we liked in the middle of the roof—away from the dead tree—where we had clear line-of-sight. This location was originally ruled out because the building had a wood roof and there were concerns that it would be difficult to establish a firm and leak-proof mounting.

Nevertheless, the antenna was moved as we asked. The problem remained. We were all scratching our heads. The link had performed well for several days and then became intermittent and eventually completely dead. We checked connections on both ends. We replaced cables, changed lightning arrestors, and swapped the radio positions. Everything looked good. Then, we performed additional testing, and it appeared that one of the radios might be defective. One end of the link could transmit and receive to the other. But the other end could only transmit and not receive.

After battling the weather and enduring shipping delays, and after several days of troubleshooting, we ordered and installed a new radio. To everyone's surprise, the problem remained. I was starting to get some heat from the project manager. The move-in date would have to be delayed. At the direction of the manufacturer, we performed more troubleshooting and testing. The final verdict was in: We were receiving interference. A new set of radios and antenna would have to be installed using a different frequency. The current radios operated at 5.3 GHz. The new radios would use 5.8 GHz.

The new equipment was installed the next week. The weather continued to slow down the process. The new equipment worked perfectly. The link was good. We decided to allow the link to run for a least a week before we gave the okay to begin the move. We felt relieved and confident that our efforts had paid off. The data link was completed, but we had one last hurdle to clear. Now that the link was up, we could bring in the voice vendor to complete its part of the installation.

The voice vendor had no prior experience with this type of install. It had designed a solution that would extend our current PBX system to the new facility via the "wayside T1" that the wireless link provided. The Tsunami equipment is designed to accept a T1 signal at one end and make it appear at the other end. The engineers at Tsunami explained it this way: "You stick a T1 in one end and you get a T1 back out the other end." That sounded simple to us, but it was not so simple for the voice vendor. After another week of troubleshooting and tweaking, the voice system was functional. The move-in date was finally scheduled.

Final analysis
Although we had quite a few tense moments during the installation and ended up delaying the move-in date by a month, we would still choose wireless if we had to do it again. Wireless can offer significant cost savings and increased bandwidth over traditional leased-line links. As we found out, wireless requires longer lead time for testing and extra planning, and it creates more potential headaches in trying to pull off a successful installation. But, if you are willing to stick it out, the cost benefits are usually quite rewarding.

With the knowledge we gained in this installation, we went on to replace the old wireless link that had connected the original location of the business unit we moved. We performed that installation ourselves using Tsunami's QuickBridge 60 wireless kit, which included everything needed to complete a link. We completed that project in a single day.

Editor's Picks