OLPC hardware: what if there was an SDR modem / chipset?
Luke Kenneth Casson Leighton
lkcl at lkcl.net
Tue Jan 26 05:21:37 EST 2010
On Tue, Jan 26, 2010 at 3:55 AM, John Watlington <wad at laptop.org> wrote:
>> _if_ all these things were true, _would_ the OLPC hardware design
>> team select such a COTS modem and its associated firmware
>> (over-and-above the rather dull Marvell 88688 option being deployed
>> right now in XO-1)?
> The XO-1 uses the 88W8388, which I've never heard anyone describe
> as "dull" before...
i mean in terms of free software, as in being user-programmable, as
in no NDA required to program it. as an overall solution which
assumes that free software user-programmability is high on the list,
then anything that includes proprietary technology, even if it's damn
good proprietary technology that otherwise shines, i would
colloquially call ... well... "dull" :)
> The XO-1.5 uses the 88W8686, which I hope is nice
> and dull from an engineering point of view (it has been so far.)
he he :)
... did you have to sign an NDA in order to get the firmware, and can
the 88W8686 be reprogrammed by anyone? if i want to write my own
peer-to-peer 802.11 algorithms, doing an implementation e.g. of the
Babel routing algorithm to run actually on the WIFI chip itself, can i
do so, right now, _without_ being forced to sign a Marvell NDA?
(babel loop-free routing protocol:
>> and if so, who do i need to talk to, to get a "yes" from?
>> (and if not, what _would_ it take for an SDR modem to be selected?)
>> this last is perhaps the key question. what's it going to take to
>> get an SDR modem into a future XO?
> Equivalent performance, for an equivalent cost and power budget.
ok. but this is for WIFI only, right? because that's the only thing
that doesn't require licensing, yes?
so this splits into two.
1) WIFI only replacement using SDR modem
* what's the current price;
* what's the current power budget (given that 802.11 is 100mW max or
400mW max in Hong Kong)
2) WIFI replacement and then some
what's the point at which the additional benefits (*) just become
sufficiently compelling that it's a no-brainer decision? $5? $8? $10?
(*: being able to do 802.22 - "White Space" Cognitive Radio Broadband,
WIMAX, GPS, GPRS, EDGE, 3G, LTE, 802.16, 802.11, FM radio, DVB or in
fact anything where you're prepared to add the extra antenna and the
> Whether to use fixed hardware or software is a frequent engineering
> decision. At the prices you quoted, I can currently get multiple dedicated
> hardware interfaces for the same price, and almost certainly for lower
i'll find out about the power requirements. i don't expect them to
be stupidly high - the RF front-end chip is 40nm after all, and
operates at 1.1v.
there _is_ a baseband CPU i heard about, but it's not being pushed
through to COTS status right now, unlike the RF chip.
> The certification issues for SDR haven't been settled, either.
> I don't know of any products w. SDR shipping with user modifiable
> software at this time. Can you point me to some ?
... you're not going to like the answer :)
* the USRP from ettus research ($750 and made of discrete components)
* the lyrtech SDR developer kit (again, discrete components)
the A2D and D2A chips from e.g. Analog Devices have insane markups
(and certainly aren't 40nm) so cost around $40 each. the FPGAs being
used to do the frequency shifting have to be huge, and are priced at
around $75. etc. etc.
that's about it, after months of searching. _apart_ from these three
chips (2 transceivers, 1 high gain low S/N receiver)
non-user-programmable (i.e. NDA, cartels etc.):
* the X-Gold SDR20 plus SMARTi SRx RF transceiver from Infineon.
designed specifically for GSM and 3G, the SDR20 also includes
GSM timing circuits, GSM encrypt/decrypt blocks, SIM card I/O
* several base-band CPUs including a 3G one from qualcomm.
(i've written off qualcomm already due to their $15 software
license fees per modem. the company is owned by 3 lawyers).
More information about the Devel