Public safety support, government agency networking, and residential Internet access are all familiar reasons to roll out a municipal wireless system. But what about voice? The mesh network that enables ubiquitous urban coverage could also support a whole host of voice applications:
Steve Rayment from BelAir Networks
walked through the variety of mesh network types that they support:
single-radio shared, multi-radio switched, and dual-radio shared.
Shared mesh time-slices the spectrum between each node, so you can
process ~ 1400 packets/second. Switched mesh fills the spectrum between
each node, so you get ~ 5000 packets/second. With switched mesh, BelAir
estimates that you can handle 4000 users per node.
This all sounds great, but it means that quality of service becomes mission-critical. Modest disruptions or changes in latency can create ripples of adverse effect. Deploying carrier-grade, voice-capable networks will be much more challenging than supporting Skype calls.
To this point: Matthew Gast from Trapeze Networks dug into the issues of QoS. VoIP over 802.11 is currently dominated by proprietary standards for security, radio management, and CAC extensions. For VoIP to work better over 802.11, 802.11 needs to work more like the cell phone network does. Applications for capacity management (in particular, a directed handoff to the next access point as you walk down the street) are wanting. To get past the myriad issues here, 802.11 hotspot managers need to stop thinking like data network managers, and start thinking like telephone network managers.
Here's to the O'Reilly Emerging Telephony Conference (eTel) for bringing these concepts together.