Differentiated Services Promote Wireless Innovation
While I don’t follow the net-neutrality debate closely, it is my understanding that the consumer advocates consider offering Differentiated Services at a premium price somehow conflicts with the Internet’s promise of equal access for all. The idea being that equal access for all, or an impartial, non-differentiated Internet, is better for everyone. I would like to turn this argument somewhat on its head firstly by showing that MNOs are already providing provider specific, Differentiated Services that regular consumers are paying for today. Secondly, that if MNOs provided general Differentiated Services to consumers, competition and innovation would accelerate, particularly in wireless network. And finally, because MNOs are not providing general Differentiated Services, they are locking consumers into their content. The good news in all of this is that unlocking Differentiated Services that MNOs can charge for is likely to increase revenue opportunities for MNOs while increasing services for consumers.
The Differentiated Service that MNOs are currently offering to consumers is Voice. While existing cellular networks do not deploy VoIP as a mainstream service, the Voice service that you get from the existing Cellular network today is indeed a differentiated service. Services are differentiated not at the IP packet level as would be the case for a VoIP service, but rather at the radio RF level where RF capacity is reserved for the purposes of carrying your voice call. When this RF capacity is reserved for Voice, it cannot be used to carry best effort IP data traffic.
I’m not proposing that we open up the 3G UMTS voice circuits or their RF capacity to alternative VoIP providers. There are other technical reasons why VoIP over UMTS is not likely to perform well. Existing UMTS data services do not lend themselves well to QoS, particularly when compared with the robust UMTS R99 voice. However, looking forward to LTE, there is an opportunity to open up the playing field. VoLTE is essentially a VoIP technology. However, it is being specified and deployed in such a way that the MNOs will still control the resources required to carry high quality voice, thereby retaining the ability to charge for voice minutes. A whitepaper from Broadsoft (http://www.broadsoft.com/pdf/volte-white-paper.pdf) summarizes the intent of the MNOs nicely:
“The adoption of VoLTE across the global MNO community is significant on a number of levels. Firstly, the price point of IP-based communications is such that MNOs will be forced to innovate in terms of how services are offered, packaged and priced. Consumers will not tolerate being charged over the odds for mobile VoIP, when compared with Over the Top (OTT) services that are nominally free outside of mobile internet data charges, unless operators can show the value within a variety of communications options.
Secondly, the Quality of Service (QoS) on connecting network-based mobile VoIP calls will be much better that OTT services, thereby creating a de-facto two-tier system. A VoLTE call would be explicitly identified in the operator domain as real time conversation and routed with low latency transit from network to network, whereas an OTT call would be treated as traditional data on the internet.”
Which begs the question, why should only MNOs be in the position to offer the upper tier of the two-tier system? It seems that by not offering Differentiated Services to customers, the MNOs are able to lock customers into their IP content, which in this case is the VoIP service, and exclude all OTT providers from being able to deploy a high quality VoIP service over the mobile network. So, the net-neutrality consumer advocates have it all backwards. If MNOs were to offer open Differentiated Services for which they could charge a fee, consumers could choose their voice provider. Once such Differentiated Services are available, it is likely that a new set of real-time, mobile applications will emerge including high fidelity voice, face to face video and real-time gaming.
Future networks will have to be engineered carefully so that real-time traffic does not starve out the best effort traffic. This can be accomplished fairly easily by allowing only a small fraction of the available RF capacity to be reserved for real-time services. SLAs can be written and network capacity planning can be done with Differentiated Services in mind as is done in today’s wired network.
The real promise of the Internet is not that it is somehow a free resource open to all. Such models always fall victim to the “Tragedy of the Commons.” The real promise of the Internet is that it is the end-points that define the applications and drive the innovation, not the network. The network is only an enabler, and Differentiated Services are an enabler, not an application. Voice is an application.
Once Differentiated Services are available from mobile networks and MNOs can charge for them, it is likely that the end-points will begin to innovate and use these resources at a much higher rate than the mere bandwidth required to carry a voice call. This, in effect, is likely to increase demand and therefore revenue for what the MNOs are supposed to provide: IP connectivity to enable applications rather than the applications themselves.
There is one “holy grail” application that could be enabled by Differentiated Services offered by MNOs, specifically Unified Communications (UC). I will define UC as the ability to provide voice and data services over a single network infrastructure, which is assumed to be IP. The dream of UC has been derailed by mobility. Mobile devices sometimes connect to the service provider network and sometimes connect to the customer’s network, usually via WiFi. This all works reasonably well for data traffic since most data applications can deal with a change of IP address and a momentary loss of connectivity. The same cannot be said for VoIP calls, which will be dropped or suffer poor performance during these transitional periods.
This problem is particularly acute for enterprise voice. Enterprise VoIP works fine over the indoor WiFi network, but when the mobile device connects to the cellular network, it cannot allocate resources to handle the voice session and then coordinate the hand-off from WiFi to cellular. Several vendors have tried to interwork with the PSTN and revert to a traditional cellular circuit switched voice call with very little success.
If mobile devices have the ability to signal their QoS requirements to the network and get feedback from the network regarding the ability of the network to continue to service their request, they could make intelligent choices about when to switch from cellular to WiFi and vice versa. Such a capability would enable the installation of the corporate dial-plan onto employees’ mobile devices thereby provide complete mobility to employees and connectivity via the network that represents the least cost and best quality for the user. This allows the WiFi network to be used when available thereby increasing voice coverage indoors without having to install expensive antenna systems or MNO provided femto-cells. The Universal Mobile Access (UMA) feature in today’s T-Mobile network is a proof-of-concept that such systems enhance wireless utility: UMA passes calls between Wi-Fi and cellular (and vice versa,) preserving QoS by making use of Wi-Fi’s 802.11e mechanism.
So let’s keep the net-neutrality debate clear with respect to the type of neutrality we’d like to demand. Neutrality should be defined in terms of applications and content. The network should be able to enable applications that require Differentiated Services so long as all the application service providers have equal access to the differentiated services. You’re not getting this today with your existing cellular voice service, but you could get it with LTE if MNOs are required to provide equal access to Differentiated Services.
One concern with allowing any kind of “fast lane” is that it introduces transaction costs where each OTT company needs to negotiate contracts with the top N different ISPs. (Unless brokers appear; perhaps CDNs could fill that role.) It’s especially problematic if innovative “two guys in a garage” startups find that the cost of negotiating access to the fast lane is a significant fraction of the cost of building their service.
For completeness, if the server can pay to get into the fast lane then the client should also be able to pay to get into the fast lane (e.g. consider an open source P2P app where there may be no company that can negotiate with the ISPs).
I see little reason for the negotiation of contracts between the OTT companies and the ISPs. The OTT provider only needs to get permission from the user to allocate resources from the provider which will be debited from the user’s account. This is similar to apps that request permission from the user to use location information to enhance the application experience.
I think we are all agreeing that transiting the core from ISP to ISP for DiffServ traffic should follow a brokering model that is quite similar to existing provider to provider voice minute brokering.
Finally, the OTT provider is billed from *its* ISP like any other user for its DiffServ usage.
Let’s just change the discussion from voice minutes to QoS bytes and everything falls out.
This sort of thing happens todayh with CDNs, as you point out. With a market for DiffServ, I imagine transit networks would serve as market makers, and consumers can do it as well, as you point out.
Realistically, we’re talking about a range of services that aren’t going to be provided by two guys in a garage.
[…] Differentiated Services Promote Wireless Innovation […]
[…] Differentiated Services Promote Wireless Innovation […]