"Best Effort Only" Network Neutrality Inhibits Many Proposed 5G Apps
It might soon be a moot issue, but some argue that common carrier regulation of consumer internet access will prevent the development of some 5G services and apps, particularly those related to internet of things.
It is hard to say, in part because of uncertainty about the boundaries between managed services and consumer internet access. The clear distinctions are between “business” internet access and “consumer” access. Business access services are allowed to offer quality of service attributes, such as packet prioritization to support some apps over others.
That is the sort of feature barred under existing “strong” forms of network neutrality affecting consumer services.
It has been less clear where the boundaries are where it comes to consumer managed services, such as video or audio subscription services or home security. And that ambiguity is likely to extend to consumer-focused IoT apps and services.
The potential use cases include self-driving cars with pre-crash sensing and mitigation, health biometric sensing and response, telemedicine, and proactive monitoring of critical physical infrastructure such as transmission lines, Rysavy Research notes.
“What many of these new applications have in common are stringent data communication requirements, such as high reliability or minimal delay,” Rysavy notes.
In other words, quality of service requirements are likely to be necessary. And that conflicts with strong forms of network neutrality that forbid packet prioritization.
In other words, might quality of service necessitate use of packet prioritization, or “fast lanes,” as some proponents of strong network neutrality have argued should be illegal?
Current consumer internet access networks assign equal priority to all third-party application traffic, regardless of the application type, as required by strong network neutrality rules that prohibit any sort of priorities between classes of apps or providers.
Traditionally, the objective of such prioritization is to maximize the quality of experience across the largest number of users and application types possible, allocating higher priority for those applications that need it while not adversely affecting those that do not.
For some IoT apps, the stringency of communications latency is even more important, not simply user experience. “At 60 miles per hour, a car travels nine feet in 100 msec versus only one inch in 1 msec,” Rysavy notes. “In a scenario of an intelligent highway warning a car of a pedestrian on the road at a blind curve, that could be the difference between life and death.”
The present uncertainty is that it is not clear whether an autonomous driving app or service is a “managed” service using the internet, or is an “internet app” used by consumers. If it is the former, an argument can be made that a method of controlling autonomous vehicles is a managed service, not subject to “best effort only” access rules.
If an autonomous car, on the other hand, is deemed to be a consumer service using “the internet,” then latency protection cannot be applied. It gets worse.
Perhaps the service provider is providing a “business” access service to the owner of the vehicle, in which case latency protection can lawfully be provided, if the owner is a fleet operator.
But does that same rule apply to a consumer owner of that vehicle?
The problem is that such uncertainties are endless, where it comes to apps and services that depend on assured latency, or assured bandwidth, to work.
To some extent, live video streaming has latency and bandwidth requirements. But is Netflix an internet app, or a managed service? What about a Comcast or AT&T streaming service? And what about health apps delivered to smartphones or other consumer health devices?
Some might argue we can create rules ad hoc, as needed. But that raises the level of business uncertainty for smaller, start-up entities, of the “it is not clear whether this is legal, or not” sort.
The point is that “business access” can lawfully provide QoS. Consumer services mostly cannot, but might, in those cases where a service is deemed to be “managed.” It is not so easy to make such determinations, and certainly not “a priori.”