A coterie of network operators got together at Spring 2006 VON to discuss growth drivers, challenges, and roadmaps around WiFi deployments (and specifically, voice over WiFi). Mike Borsetti moderated, and panelists included:
Weiss segments VoWiFi (voice over wifi) by service provider, location, and/or platform. Jonker, on the other hand, segments the industry by service provider, aggregator, hot spot operator, and venue.
Fjogstad sees that the Internet has fundamentally changed how network components relate to each other. In the "new world," any type of packet gets equal service, so long as there is an authorized user. The separation of access, transport, and service is growing, with a wide set of implications.
The consumer's constant search for cheaper, better service drives change. Consumers see and adopt wireless IP connections as just another point where they can connect, and don't want the complexity of a separate service that they must subscribe to. Growth in broadband penetration, wifi handset revenue, and VoIP competition are all behind increased awareness.
Fjogstad stressed that the business implications of IP going wireless should be dramatic. In VoIP, the distinction between fixed and wireless voice goes away; and more interestingly, the consumer will soon discover that VoIP can be used as more than a POTS replacement.
Weiss believes that the commoditization of legacy network infrastructures turns the carriers into mere transport. As a result, the carriers are converging, and providing multiple access technologies that broaden their service offering and reduce churn from their network. As an access technology, WiFi acccelerates this trend. Mobile operators, on the other hand, are entering the market in order to provide better coverage within the home, reduce churn, and defend against alternative VoIP providers. New, fixed-line operators may see VoWiFi as a way to leverage fixed-mobile substitution, or enhance a bundled service offering.
The first generation of consumer WiFi phones is pretty limited - they're still expensive, the battery life is still short, and they don't have features now common to an everday telephone. And from the vendor's side, the current wireless IP market is messy. Multiple radios increase costs, reduce battery life, hurt roaming capabilities, and reduce QoS.
Weiss' list of challenges include both security and voice quality issues - unlicensed spectrum, collisions in the MAC (or physical) layer, and backhaul bottlenecks - that result in packet loss or delays. Jonker worries about user experience and provisioning. And Jonker sees Boingo's biggest hurdle to adoption in that it's not available everywhere - more access technologies, more widely available, would really help the business.
Jonker launched into his talk with a clear delineation between his role as a network aggregator vs. a network operator. Boingo's goal is to provide consistent experience, billing, and brand for the user, regardless of their physical location. This implies that hotspot requirements include:
- Symmetrical VoIP traffic (not ADSL)
- WiFi QoS
- Support for a VoIP-capable subset of a more general network
Fjogstad predicts that wireless IP will force mobile carriers to transform themselves into wireless ISPs. Separation of access and service/application will become the natural order, and the wireless IP vendors will adapt to it. Best-of-breed ISPs will add value via access roaming. Cellular carriers are likely to build walled gardens that reject this natural order, but this will harm their competitive position.
Jonker wants to manage WiFi access through software - automatic login and registration (if it makes sense), a uniform experience regardless of location, and a small footprint on the device. This sounds suspiciously similar to what the Feeva guys are up to.
Some solutions that Weiss looks to:
- 802.11e: Supports quality of service, unlike existing WLAN standards
- 802.11n: Increases speed
- Voice layer improvements: RTP layer, multi-vocoder support, RTP bandwitch policy management, dynamic vocoder changes, and sample size management
(Does this remind anyone else of the early days of Ethernet..?)
Not to mention - what's a friendly charging model for VoIP/data? Jonker sees current models that are based on connections. For laptops, this means usage, but for VoIP devices, this means registration, standby AND usage. Jonker's ideal model would instead base fees upon actual usage - per minute for voice, or per byte for data.
- Fierce competition is driving operator convergence and service bundling
- Consumer cost savings are driving growth
- IP is as dependent upon deregulation as it is upon technology
- Economic drivers favor a single standard
There will be many points of access to IP, and many access technologies, but from the consumer's perspective, it's all just access. Once the market reflects this viewpoint, adoption is much more likely.