Air-5 and Magis Networks

In late 1999, I co-founded Magis Networks, a venture-backed start-up company focused on high quality of service (QoS) wireless networking in the unlicensed 5 GHz band. We perfected wireless distribution of high-value content like HDTV using OFDM technology. I was the company’s first vice-president of engineering, later stepping over into the CTO and Phy Director roles.

Magis Networks Company Photo

There was a lot going on at Magis as we set out to develop a new communication standard that was capable of supporting high QoS wireless networking. There were only two wireless standards at the time that were vying for dominance in the 5 GHz UNII bands; Hiperlan-2 and 802.11a. The Hiperlan-2 standard had certain benefits compared to the 802.11a standard, but was considered too bulky (and therefore costly) for a consumer product. Neither of the standards could reliably deliver the QoS that was needed to support HDTV-quality MPEG-2 distribution, however. We consequently had to venture out where no one had gone before.

The Magis Networks Headquarters

We had tremendous pressure from our investors not to venture too far away from the 802.11a standard. They insisted that we be “interoperable” with 802.11a. This stance took more sophisticated directions like MIMO off of the table. We ultimately took the 802.11a standard and augmented the preamble with some special signaling that allowed us to support multiple receive antennas in a nearly optimal manner thereby improving link QoS dramatically…probably on the order of a thousand-fold or more. This improvement in link reliability was mandatory if we were to be able to support MPEG-2 signaling without exceptionally large ARQ buffers. The signaling modifications to the 802.11a standard were the subject of several patents including US Pat 6,879,840, US Pat 6,633,616, US Pat 6,549,583, as well as other US patent applications including US 2002/0160737, US 2004/0037366, and numerous trade-secrets.

We did a lot of rather creative things to keep the ship afloat during those heady start-up days. If you found a big fish sitting in your office chair some morning at Magis, you knew that you had become the pace-setter in our fast-paced development schedule and you needed to pick up speed. When we were nearing the end of our first round of financing and we were way behind schedule, we had to do something more. We knew that we would have to have some kind of working hardware to show off or our financial valuation or the second-round of financing would be dismal. So, after Thanksgiving in 2000, I launched what was supposed to be the “45 days of Magis (crunch-time)” but it ended up being more like the “100 days of Magis.” We were building an entire system, RF, DSP, PHY, MAC, software, and we were all behind. Rather than lean on the director-level guys to lean on their people to work longer and harder, I took a different approach. I required that everyone work late (with dinner provided) on Tuesdays and Thursday nights. The other nights of the week, we really tried to encourage people to go home at a normal time. On Valentine’s day, I almost physically pushed people out of the building before 6 PM to go home to their spouses. The end-results were impressive; we made a working hardware demonstration for our investors by the skin of our teeth (like one hour). During those difficult days, everyone knew how high to jump, and kids and spouses knew what nights they could expect to have everyone home at an early hour. The loyalty that this created in the design team to the tasks at hand and to one another was nothing like I had ever seen before. And perhaps as important as anything, the sanity of this approach resulted in no casualties and serious burn-out.

Yagi Array

Some of the technical results of our efforts are recorded in the documents in the right sidebar. Wireless technology has moved forward dramatically since this technical work was done, in part fueled by new algorithms and methods (e.g., MIMO), but also by tremendous advances in semiconductor technology that makes everything possible.