Which way WiFi?
WiFi users in densely populated urban locales experience signal interference from nearby unassociated WiFi networks. Additionally, significant growth in the number WiFi-enabled mobile devices has added to user dependency on WiFi channels and to the data traffic those channels carry.
Early implementations of WiFi networks developed on university and corporate campuses, which share a number of characteristics and interests: Both tend to occupy large areas demanding multiple access points to serve large user populations. Both tend to have perimeter space separating the occupied portions of the campus from neighbors, so interference with and from adjacent locales is rarely a concern. Suburban residential areas similarly often feature sufficient space between homes so that network-to-network interference is rarely problematic.
Large, multi-AP (access-point) campus installations operate with a significant advantage over densely situated APs belonging to independent networks: A central IT department uses tessellation maps to assign channels to campus-based APs, allowing adequate spacing to prevent interference between APs operating on a common channel.
In urban settings, no such coordinating authority exists and the unit-to-unit pitch in apartment buildings is usually quite small. Individually owned APs, therefore, tend to interfere with each other, reducing the attainable data throughput rates for all.
Part of the problem is that the vast majority of APs operate in the 2.4-GHz band. Although that band provides 13 channels (11 in North America), those channels are only spaced 5 MHz apart. WiFi APs use 22 MHz of spectrum. With such bandwidth allocation, there are only three channels with no spectral overlap: Channels 1, 6, and 11 (Figure 1). Newer, very-high-speed APs operate with 44-MHz-wide links, exacerbating the challenges of an already difficult RF environment.
Areas that use minimally overlapping channels — 1, 5, 9, and 11 — suffer sufficient mutual interference to reduce data throughput rates by nearly half. Most APs suitable for SOHO (small office / home office) or residential use provide for automatic channel selection based on detectable spectrum use. Still, surveys of various locales including urban settings, suburban commercial districts, and residential areas confirm that AP channel assignments are a hodgepodge with little evidence of channel selections that avoid mutual interference.
There is also little evidence that, outside of large institutional settings, APs make use of the channels in the 5 GHz band. Trial surveys JAS Technical Media recently made in New England and upstate New York show only a few percent of APs operating in the 5-GHz band. Greater adoption of 5-GHz APs could give much needed relief from spectral crowding in the 2.4-GHz band, particularly in densely populated areas.
The trend among AP equipment manufacturers has been to stress larger link distances and faster maximum data throughput rates — both of which can lead to greater mutual interference. In a market that reads greater range and speed as better performance, this trend may be counterproductive for many consumers: In a crowded room, everyone shouting makes conversation harder, not easier.
Another option that occasionally receives attention is LiFi — optical communication links, which solve the interference problem owing to the convenient fact that light doesn’t pass through most walls, floors, or ceilings as RF so readily does. The first mention of LiFi development I’ve found dates back to 1970 at Boston University, so the idea isn’t new. It has found research interest around the world with projects in this decade at Pennsylvania State University, the Heinrich Hertz Institute in Berlin, EPSRC (the Engineering and Physical Science Research Council — a consortium of UK colleges led by the University of Strathclyde in Glasgow) and, most recently, Fudan University in Shanghai.
The EPSRC team is developing a multi-segment LED transmitter, dividing the area of a 1-mm square LED into as many as 1000 individual emitters. The Heinrich Hertz Institute team has demonstrated a 800 Mpbs downlink with a 90 square foot range. Both of these projects appear to target institutional applications.
This past October, the Fudan University team reported a 150 Mbps downlink using a 1-W LED. This implementation, though less technologically advanced than others, may also be more easily commercialized at price points attractive to residential, SOHO, and retail locations.
One challenge to LiFi that seems to escape most discussions is the uplink path. Little information is available about uplink performance, client-side transmitter shadowing, or AP-side receiver criteria. Simple data-transfer requests can succeed on low-bandwidth uplinks because the data density is substantially sparser than that of the downlink. However, increasing use of IP telephony and other real-time communication that depend on higher uplink bandwidth suggest link bandwidths that are closer to parity than was necessary in years past. Additionally, LiFi may face market resistance unless rollouts include pad computers and smartphones at roughly the same time as adapters for laptop and desktop computers.