A History of Home Networking Technologies (pt.1 wireless)
I've been part of the effort to make home networking take off for the last seven years and I've seen many technologies come and go. In this posting, I want to take a trip down memory lane and talk about the physical home networking technologies from the past, where we are today, and what I think looks promising for the future. This issue, I will start with wireless.
I started into Home Networking as part of the "Anywhere in the Home" Initiative at Intel back in 1998. Back then there were various efforts to solve the "barriers to entry" for Home Networking. When people talked about Broadband, they were talking about ISDN and DSL was just coming onto the stage.
The wireless ethernet technology of the day was FHSS (frequency hoping spread spectrum) that was dominated by Proxim and was pretty much proprietary to them. It operated in the unlicensed 2.4 GHz range, but delivered 2 Mbps (which means you got around 1 Mbps). There was a new technology called DSSS (Direct Sequence Spread Spectrum) that was being proposed by Lucent and Harris Semiconductor (later spinning out their wireless division into Intersil) that delivered about 2 Mbps as well. DSSS would later become the Wi-Fi and 802.11b/g that we know today. Both technologies were run through the IEEE as 802.11, causing confusion when talking about 802.11 because there were two flavors, frequency hopping and direct sequence.
My early analysis of both technologies revealed that DSSS was superior to FHSS in many ways. FHSS would interfere with all the other devices in the unlicensed 2.4Ghz spectrum because it would transmit up to 1 Watt (FCC max) at every hop (probably 1 Mhz) between 2.4-2.483Ghz range. I had been using wireless video senders from X10 to send video around my house and these FHSS radios instantly created snow on my TVs. The DSSS radios transmit on a 22 Mhz wide channel that has a peak around .1 Watt. The nice thing about this was that I was able set the channels on the DSSS radios to not overlap my wireless video senders and everything in my house would place nice. This meant that DSSS was more interoperable with existing devices (baby monitors, video senders) than FHSS. The other nice thing about DSSS was that the bandwidth for it was about to go to 11 Mbps (divide by 2 to approximate TCP bandwidth) with the 802.11b spec.
Some folks at Intel had the grand plan to introduce a new standard called HomeRF to make wireless home networking a reality for new usage models. HomeRF was a combination of FHSS and DECT (a European cordless telephone standard). HomeRF was a product of Intel Architecture Labs and was a novel implementation of the wireless technologies of the day by some clever engineers. However it ignored what the market needed, so was doomed from the start. HomeRF was to provide both a dynamic asynchronous (for data) and 0-4 synchronous (for voice) channels for the home.
The theory was that HomeRF would provide a single wireless technology for the home network. The engineers at Intel reasoned that the volume of cordless phones would help drive down the cost of the wireless technology. Siemens was a big proponent at the time because this was their way to enter the US market which was being dominated by Panasonic and Sony cordless telephones.
There were a lot of problems with HomeRF that were overlooked, and this ultimately led to its downfall. Homes didn't typically have more than one voiceline, so having the ability to have 4 voicelines was unnecessary. The idea of combining two different technologies, FH and DECT, also formalized interference between the two usage models, so there was a bigger degradation in the asynchronous channel, than if the technologies had just operated independently of each other and interfered with each other. If for some reason all 4 voicelines were being used, the asynchronous channel would be reduced to nothing. Definitely not the idea! The volumes never materialized for HomeRF either to make it cheaper. FHSS was cheaper to make at the time than DSSS, but that was only temporary. I could go on and on about HomeRF's shortcomings, but maybe because I couldn't steer Intel away from this crash course, I'm just a little bitter, so I'll quit my rant now and move on.
So, I had proposed that HomeRF consider the DSSS technology, but Intel was already in bed with Proxim and eventually invested $10+ million with them. I moved on and helped Intersil succeed with a better version of wireless networking that would eventually become known as Wi-Fi and 802.11b. I helped some fellows from Intel's mobile division start a wireless division for "business" that would use the Intersil (DSSS) technology. At the time, the DSSS was a 7 chip solution (expensive), but there was a roadmap in place to take it to 4 chips, then 2, and finally one chip. Businesses were starting to adopt wireless networking which helped drive the volume up and price down on DSSS. Wi-Fi was a marketing organization to promote this technology and was very effective.
Around this time, another technology called Bluetooth was being proposed, but it's main purpose was for cable replacement and was low power/short range. There were threats by it to be a longer range networking technology by boosting it's power, but Bluetooth is a point-to-point networking technology based on FHSS and never really made a run at home networking.
Intel, missing the boat on 802.11b, then tried to get a jump on the game by promoting 802.11a (unlicensed 5 Ghz). 802.11a used a new protocol called orthogonal-frequency division multiplexing (OFDM) that was hard to do up until then, because there wasn't enough processing power on chips. 802.11a OFDM has 52 parallel sub-carriers it runs and claimed 54 Mbps (divide by 2 for approximate TCP throughput).
"Since the 2.4 GHz band is heavily used, using the 5 GHz band gives 802.11a the advantage of less interference. However, this high carrier frequency also brings disadvantages. It restricts the use of 802.11a to almost line of sight, necessitating the use of more access points; it also means that 802.11a cannot penetrate as far as 802.11b since it is absorbed more readily, other things (such as power) being equal."
With this limitation, 802.11a was not a good technology for the Home Networking market, because consumers don't want to buy 4 access points for their house. 802.11a did have its success though in the office/cubicle environment where you'll often see the access points mounted on the ceilings to clear over all the cubicle walls.
802.11g was ratified in 2003 and provided 54 Mbps, which matched 802.11a. 802.11g was backwards compatible with 802.11b, so it was attractive for people that had already invested in a 802.11b network. The problem with 802.11g is that it can't run at full speed when the network has a mixture of 802.11b devices on it. The high bandwidth also can only be achieved with good signal strength and low interference, otherwise it behaves like 802.11b.
802.11n is a future wireless networking technology that incorporates a new technology called MIMO (multiple input, multiple output) that increase theoretical bandwidth to 540 Mbps. It is slated to be a standard in 2006, but many manufacturers are shipping early versions to get a jump on the market. This may be a hot product for the Holiday Season 2005.