Wireless Mesh Reality Check: Home CPE Required

A blog entry by Michael Sciannamea in The Wireless Report caught my eye late last week. The post summarized an article concerning coverage in Tempe, Arizona's muni-wireless mesh network, and ended by saying "Hopefully, these bugs will be worked out quickly..." If providing adequate coverage were only a bug! A comment on that post by fellow colleague Glenn Fleishman pointed to an article that was the likely original news source. In that article, NeoReach, the wireless network operator, admitted that it had not fully informed registering customers that the wireless service generally required an indoor bridge for the best possible experience.

Requiring end-user CPE for broadband service is not new: DSL and cable broadband subscribers have always needed a modem, and most BWA (broadband wireless access deployments) deployments have generally required outdoor-mounted antennas. What's disturbing is that the image engendered by mesh network operators or vendors in their presentations and marketing literature to city officials is that subscribers would be able to use their Wi-Fi enabled devices everywhere. Unfortunately, that panacea has not become a reality in the home. You may fantasize about using your laptop to check email or work on a report from your backyard patio's lounge chair, but as I look up through my basement window the thermometer registers a cool 40 degrees. While many wireless trials target the downtown core with street-facing cafes or open-space areas, most residential consumers access the internet from inside their home or apartment where wireless coverage will be poor.

When wireless mesh nodes are deployed 20 to 40 per square mile, back of the envelope calculators show that each node covers a radius of 450 to 650 feet. Tempe's deployment, at last count, has 550 nodes over 40 square miles, or an average of 14 nodes per square mile which means a radius of about 800 feet. Typical, line of sight distances for SOHO wireless access points at full link rate hovers around 300 feet. Even if the mesh vendor increases the RF power output of the radio and receiver sensitivity, it's should be pretty clear that extended distance, outdoor foliage, building materials, and multi-dwelling unit design will almost completely attenuate, or reduce, the signal to most inner rooms. Neoreach's own website for the Tempe deployment says that "Individuals in the deployed zones may be able to connect to the network indoor if they are within 150 yds [450 feet] of an AP, others may need additional equipment to enhance indoor coverage." To get around the distance problem, a wireless bridge is placed on or near an outside window facing the closest mesh node. A wired port feeds the home network or a local computer, or perhaps a wireless access point/router, hopefully on a different, non-overlapping channel.

Several challenges arise from requiring home CPE. The first and most obvious one is that wireless mesh provider must educate their subscribers that they need to purchase a wireless bridge. This entails a cost the subscriber wasn't likely expecting (don't I already have Wi-Fi support on my laptop?), guidelines for correct placement, and maps so that the customer can identify the closest mesh node and position the home CPE correctly. Second, if the customer was looking for wireless service in their own home they will need to install and configure a wireless access point. Third, if wireless access point defaults to the same channel as the mesh radio's access service co-channel interference will ensue which in turn reduces both local and mesh node wireless performance. While some mesh vendors have promised channel agility in upcoming releases, it's more likely that the wireless mesh providers will standardize on a certain channel allocation scheme and ask that their customer use the remaining two 802.11b/g channels in their home Wi-Fi network.

Mesh customers also need to set their expectations properly concerning wireless mesh performance. To keep costs down, many wireless mesh deployments have chosen models that host a single radio for both backhaul and access, something market leader Tropos Networks offers. Other vendors such as BelAir, Cisco, newcomer GO Networks, Motorola, SkyPilot and Strix have shied away from this kind of compromise. Most wireless mesh solutions promise throughput rates around 1 Mbps, and some promise T-1 (1.5 Mbps) equivalent service, although I'm skeptical that they can offer a carrier-grade replacement to wireline T-1 circuits.

But a Wi-Fi-based solution does have a few things going for it. Unlike DSL and cable modem service, Wi-Fi is a fully symmetric service, but the broadband plans I've seen show the muni-wireless folks mirroring the incumbents' asymmetric offers. It's also true that in a controlled, short-range scenario could offer sustained rates over 25 Mbps, but the reality is that in a mesh deployment those speeds will not be attained because it's a shared service suffering from signal attenuation and sporadic in-building interferers such as cordless phones, microwaves, cordless mice. At the end of the day, marketable throughput levels will fall below traditional wireline incumbent operators which range from 1.5 (DSL) to 6 Mbps (cable) downstream.

I haven't touched on concerns related to security, the digital divide, or how operator plan on accommodating new wireless standards, but as city-sponsored wireless networks trials move from serving public parks and other open recreational areas to public/private partnerships in a city-wide deployment, consumers should expect to purchase a wireless bridge. As my colleague Peter Rysavy quipped in a recent email, perhaps wireless network operators will need to make clear whether the product is for indoor or outdoor coverage. Welcome home, and keep the bugs out.