Wed, 04 May 2005

Current State of Bluetooth

As much as I do not like to admit, Bluetooth is pretty neat. Its painfully slow FHSS physical layer is shadowed by its nearly universal vendor support, object exchange, serial cable replacement and digital voice capability.

While Bluetooth was great for GPRS, and perhaps 1xRTT and EDGE, its turning into a bottleneck. Currently, WCDMA/UMTS and EVDO speeds far exceed the maximum symmetrical serial rate Bluetooth can deliver. What does this mean? Break out your USB cables again folks! Bluetooth is now useless.

But why did the Bluetooth SIG let an aging physical layer linger for so long? Especially when standards-based technologies like 802.11/WiFi have seen profound increases in physical layer speeds. While there are many political reasons, the adoption of a non-standards based unscalable FHSS technology did not help the issue.

What was the Bluetooth SIG thinking? The Bluetooth vision was an application, not a set of physical and electrical specifications. Yet, development went in to create a completely proprietary radio interface and protocol that barely plays friendly with other spectrum users.

This likely reminds readers of the HomeRF days, yet another proprietary wireless LAN protocol competing with an open standards based specification. HomeRF is dead; Bluetooth is dying.

But wait a minute, Bluetooth is useful! What about the cool headsets, ease of swapping contact information, and multiplayer java games? Why can't we take Bluetooth to the higher OSI layers, like any other real application, and use a mass produced, well supported, nearly universal physical and MAC layer like 802.11/WiFi?

Blueooth over WiFi??

Now guys, I'm not on drugs. I am talking about the Bluetooth application suite on top of WiFi instead of its old FHSS counterpart.

Some advocates say Bluetooth's advantages are in footprint, cost, and power consumption; I believe that this can be changed. Specific on-chip integration of a bluetooth-like stack, mass production, and reduced power output can all mitigate these factors.

While it would be great to see Bluetooth riding on IP/UDP, making it routable, it may not be very possible. Issues such as IP addressing, security, and implementation complexity will arise. I could see Bluetooth existing as its own lightweight layer 3 protocol.

Both peer-to-peer and infrastructure modes are well supported in 802.11. Concepts such as piconets and pairing could be implemented with ease. Devices could easily attach to WiFi networks and packetized voice services would operate nicely. Bluetooth would not rely on functions such as WEP, as better encryption facilities could be built into the higher layers.

As WiFi advances, Bluetooth over WiFi could advance. Much like the original 802.11 DSSS cards operating in concert with 802.11g radios. It will be a long time before even the 802.11b MAC becomes too slow to be usable.

Will we ever see Bluetooth over WiFi? Most likely not. I suspect the Bluetooth SIG has the same amount of intellectual property interest in the MAC as they do the actual bluetooth application. Besides, if you could do it over WiFi, you really wouldn't need a Bluetooth chip anymore. And that means less property control and royalties.

Will we see Bluetooth Turbo? Wireless USB? UWB? The Bluetooth SIG can sit on this topic forever. The wireless industry will not. Micro WiFi access points are already being considered to correct the bottleneck in next generation EVDO/UMTS phones. If somebody invents a Bluetooth-like application protocol, the party is over for old 10th century Danish King Harald Blatand; old Bluetooth himself.

[/wifi] permanent link