Are you sure that 802.11b will use less power? I know that with Bluetooth 2.0, I've read that the power consumption per second (millisecond? of transmission time is comparable to Bluetooth 1.2, but 2.0 is about three times as fast as 1.2 - thus, the net effect is that for a certain amount of data, 2.0 will use less power.

So, I honestly don't know if you'll save power by switching to b instead of g, but I'd expect that you wouldn't find any savings, and might even use more power. Most of the g chipset vendors claim that their g chipsets save power compared to their b ones as well, but that's not the same as this issue.

From the document: http://focus.ti.com/pdfs/vf/bband/80211_wp_lowpower.pdf (page 8):

Quote from that document:
"While 802.11b may consume less power at any instant in time, the length of time needed to transmit/receive a meaningful amount of application data can be five times longer on an 802.11b network than on an 802.11a/g WLAN.

According to this 802.11b uses less power, while transmitting data below 11Mbs. If surfing web with typical DSL line (=below 11Mbs) 802.11g is overkill and uses more power.

Have to add this :
From the document:
"In a typical WLAN with shorter 802.11a/g packets, the power consumption advantage of 802.11a/g over 802.11b increases to more than three times (3x).

So if 802.11g (54Mbps) is about five times faster than 802.11b (11Mbps), it uses 5 / 3 = 1.67 times more energy than 802.11b ?
Disclaimer: I am far from being an expert on this...

Excellent document! Thank you, Titus.

The bottom line from that document is that 802.11g is more efficient in a number of ways, and in most situations, transmitting the same quantity of data will use less total power with 802.11g than with 802.11b.

The graph on the bottom of page 8 is excellent - it shows you how much power you use per bit at various bit rates.

You mention surfing with a DSL line, and indicate that that's slower than 11Mbps normally.

The speed between your DSL router and the phone company's equipment is, indeed, less than 11Mbps. However, when you're online using wifi to a DSL router, you're not talking _directly_ to the phone company hardware. Rather, you send a packet of data from your computer to the DSL router, which then sends the packet over the phone line to the phone company.

This is done an _entire_ packet at a time. Your computer can send the packet to the DSL router as fast as it wants; once the DSL router has it, it'll send it slowly over your phone line to the other end. Conversely, when there's a packet for you, it'll come slowly over the phone line. Once it hits the DSL router, the router will send the entire packet to your computer as fast as it can - which for 802.11g will be 54Mbps.

Your DSL router can store some packets in memory - so if you send too many for a little while, it'll store them and send them out as quickly as it can. Pretty soon, though, it'll run out of space and just drop some of your packets if you keep sending data too quickly.

It's really very normal, however, to have links of various speeds inbetween two computers, and to not actually be entirely sure how fast the slowest one is. You just send your data out and let TCP handle the rest

Thanks Bhima for your explanations. I also should have read the linked document a little bit more carefully, not to just jump to the juicy page 8, and make my own conclusions.
For other hasty manual dodgers interested in subject (sorry for long quote, but this and Bhima's comments pretty much concludes my original question):

"In theory, it would appear that the simpler, lower throughput 802.11b modulation scheme would result in lower battery power consumption. If one only examines the power consumed to transmit or receive a byte of data, for example, then an 802.11b device would consume approximately 30 percent less power than an equivalent 802.11a/g device for that same amount of data. But it can be very deceptive if the analysis is limited to an examination of solely the per bit or per byte power consumption when the device is in an active mode transmitting or receiving data."

And:

"Especially when 802.11 packets are short, overhead can dominate 802.11 traffic. When the standard IPv4 protocol is implemented, roughly 45 percent of the WLAN's traffic can be attributed to 802.11 MAC encapsulation overhead. Furthermore, 802.11b has a header that is strictly overhead and it is, at a minimum, five times longer than the 802.11a/g header. The following section shows that lower data rates and longer headers cause grossly high power consumption for 802.11b WLANs relative to the power consumption of 802.11a/g networks."

Have to add this to this old thread concerning differences between 100mW and 10mW transmission setting:


Source: https://maemo.org/bugzilla/show_bug.cgi?id=256

"The option of choosing the 10 mW transmission power is there only because
certain countries require this in their law. The low power transmission has a
negligible effect on the battery lifetime because the data is sent in short
bursts that last for less than 1% of total time. On the other hand, the lower
transmission power will make it harder to keep connected. Therefore it is best
to use the (default) 100 mW power, unless legislated otherwise in the law of
your country."