Why The Global Digital Divide Persists

Figure 1. Quantifying the world’s massive Digital Divide.

There’s a 2-part answer to that question. 

The first part is straightforward: installing or upgrading wired infrastructure in The Last Mile in order to deliver fast connectivity is a slow and expensive process. It involves a lot of old-fashioned manual labor (where Moore’s Law can’t help) digging trenches for fiber optic cables or stringing them on poles — and yielding unattractive returns on that investment outside of high-density markets. Fiber installations do nonetheless continue to occur, but not at a rate that closes the global divide in a meaningful timeframe, or ever in the hinterlands without untenably massive subsidies.

The second part, regarding the potential role of wireless technology in closing the divide, is more complicated. It involves a set of harsh realities in the fixed broadband application context that have significant implications for the design of a wireless approach to the problem in mainstream markets.

Fixed Wireless Access Realities

The first and most important harsh reality in fixed wireless access (FWA) is pervasive obstructions. As anyone familiar with radio design in the mobile wireless world attests, the airwaves between a cell tower and a residence in mainstream markets are full of obstructions. As Figure 2 shows, the probability of having a clear line-of-sight between a tower and a suburban home drops precipitously as a function of distance. Beyond a couple hundred meters, more than 80% of target homes are obstructed by trees or other homes. These common non-line-of-sight (NLoS) conditions have a number of significant implications that drive the majority of the essential ingredients we outline in the next section.

Figure 2. The scarcity of line-of-sight from towers to homes.

The second reality is the huge difference between fixed and mobile network traffic economics. A typical mobile subscriber consumes something like 10 or 20 GB of Internet in a month, while a typical household consumes on the order of a terabyte per month (i.e., ~50x the total tonnage of a mobile subscriber) while paying a similar ARPU.

The third fundamental is demand density. As the census tracts sample in Figure 3 shows, an FWA system deployed by an operator aiming for any reasonable market share must be able to support hundreds of households per square km.

Figure 3. Typical example of US broadband service options as a function of census tract household density and median income.

The final fundamental reality is incumbent response. In markets with a monopoly provider, a new entrant can be closed out of the market through incumbent pricing action — a common occurrence in the US. (Note the prevalence of single-provider census tracts in Figure 3).

From Realities to Requirements

A number of interrelated implications for system design follow directly from these 4 inherent realities in the mainstream-market FWA application context.

Fixed line-of-sight (LoS) radio links, with very narrow-beam antennas pointed precisely at each other, are difficult to install because of that precise alignment requirement, but they are relatively simple to design and manufacture. Radios that can successfully work around obstructions are exactly the opposite: a challenge to design but easy to install, since their alignment is all driven by software. The not-small design challenge for these radios involves meeting the following 7 requirements all at the same time.

InvisiLight® Solution for Deploying Fiber

Go to Market Faster. Speed up Network Deployment

Episode 10: Fiber Optic Closure Specs Explained…

Food for Thought from Our 2022 ICT Visionaries

7 Key Ingredient Requirements

1. Active self-interference avoidance
2. Precise multipath integration
3. Rapid adaptation
4. Continuous asynchronous interference cancellation
5. High capacity (spectral efficiency)
6. Tailored Ethernet management
7. Long NLoS range

1. Active self-interference avoidance: As Figure 4 illustrates, NLoS radios use multiple signal paths between a base station and a subscriber’s home, tapping the phenomena of diffraction (bending waves around corners) and reflection (bouncing waves off surfaces). To take advantage of multiple paths, they need to have wide antenna apertures — just like mobile network cell towers. The requirement to deploy these wide-open NLoS radios in a network, to cover residential neighborhoods at metro-area scale, yields our first key ingredient: active self-interference avoidance. This must be done sufficiently well on both ends of each link to ensure uniform service delivery at the edges of cells (a goal unmet by mobile network architectures today, but quite critical for FWA, because people can’t just drive their houses out of permanent patches of poor coverage caused by intercell interference.) 

Figure 4. How NLoS Radios Work Around Obstructions.

2. Precise multipath integration: On the receive end of the link, since the multiple signal paths involve radio waves traveling different distances, they arrive out of synch with each other, creating what would be essentially just noise to a simple radio with one antenna. Each NLoS receiver must have enough "degrees of freedom" (i.e., multiple independent radio chains) to support real-time digital signal processing that can "untangle" the multiple signals from each other and precisely realign their timing so they can be summed into a single clear signal — referred to in the trade as multipath integration.

3. Rapid adaptation: Harvesting multipath to work around obstacles has another intrinsic challenge, and that is rapid change. Trees waving in the breeze and cars driving by on the street, as just a couple tangible examples, change the structure of the channels between radios, so the system must adapt rapidly to these changes. In this application rapidly means thousands of times per second.

4. Continuous asynchronous interference cancellation: The radio wave propagation behaviors NLoS radios use to work around obstacles vary in utility as a function of frequency. Specifically, reflections and diffraction can be leveraged effectively and efficiently only in the spectrum range from approximately 1 to 7 GHz. This spectrum is full of public and private mobile applications for exactly that reason, and in populous areas is quite expensive to acquire (as periodic multi-billion-dollar mobile spectrum auctions around the world attest).

Because of the ~50x ARPU-per- MB-delivered difference between mobile and fixed services, in mainstream markets with even modest household density, using expensive licensed spectrum for mobile applications always takes a clear priority over fixed service. So, aspiring FWA operators are generally forced to attempt to use unlicensed spectrum in mainstream markets. This leads directly to the next key ingredient: deeply effective continuous asynchronous interference cancellation. The rules for outdoor operation in 5 GHz dictate transmit power limitations, but little else, and as a result these channels are full of random interference that can fill up the wide antenna aperture of an NLoS radio quickly, rendering it useless. The solution is a radio-wave equivalent of a good set of noise cancelling headphones that works continuously and in real time to extract the desired NLoS signal from all the unlicensed-band noise.

5. High capacity (spectral efficiency): A now-clear 80 MHz channel in 5 GHz is a good start, but the next factor that needs to be addressed is capacity. To serve the broadband speeds everyone wants, with meaningful share in a mainstream market, requires a system that operates with very high spectral efficiency — ideally a multiple-fold improvement over what 5G mid-band solutions deliver today.

6. Tailored Ethernet management: To that end, FPGA-based systems-on-chip that make use of parallel processing are necessary for applying tailored packet processing and traffic management firmware to the broadband access challenge. That traffic management must include tuning for over-the-air congestion control, to maximize capacity and efficiency between the base radio and multiple on-premises receivers in a very dynamic physical and digital demand environment. Moreover, the FPGA System on a Chip (SoC) must integrate Internet Protocol Security (IPSec) encryption of the tunnel between the radio and the rest of the network to maintain a secure, private connection for data transmission.

FPGAs are also an ideal platform for FWA because of their programmability, even after they have been field deployed. This enables the unit to be improved from version to version without needing to physically replace components, but rather by simple software upgrade.

7. Long NLoS range: The final key ingredient is long NLoS range. This enables rapid and broad entry into mainstream markets, by covering complete metro areas from a small number of towers and precludes sustained pricing action by incumbents.

Figure 5.

We summarize the connections between the inherent realities and these key-ingredient requirements in Figure 5.

Article coauthored by Steven Glapa and Brian Klaff

Follow Ethernity on Twitter @EthernityN.