OSNews

Eugina.... With all of those devices, you will need massive up-front costs for FCC testing. Also, I would recommend an AC Adapter or an ON / OFF switch for every type of network because with ALL of them ON all the time, constantly polling to see if new wireless connnections are availble, you will be draining your battery life for no good reason.

Also, 802.11x takes significantly more power than most people realize, ESPECIALLY IF IT'S OPERATING AS A GATEWAY. Like 0.5Amps.

Blue Tooth and 802.11b,g use the same frequency range, so their antennas would be the same size, just operating on different channels. The radio hardware would have to be pretty clean and have EXCELLENT FILTERING between channels (not the cheapola stuff you see for $9.99 at CompUSA.)

The processor wouldn't have to be any big deal since it's just forwarding data and "discovering" MAC ids and network connections to make.

You'd want SOME security, ATLEAST a password on the thing, since if I'm wardriving, I can now bridge over to your PDA and read your address book, sensitive information you capture at meetings (like the drawings you make of your boss while he's speaking,) and your contacts.

For V.90 modems, there are already MANY whole solutions in FPGA / ASIC. If your product was VERY cost senstive, I'd go for the win-modem type and do as much as possible in firmware / software.

What I would recommend, if you had the resources, BIG IF,
is do all of your hardware design in a very large and very fast FPGA. Buy a Strong ARM core and implement it. There's your processor. Design in enough ram, etc. Allow for an update system via USB or other HARDWIRED connection.

When newer standards or better compression algorithms or more channels of the 2.4GHz band get opened by the FCC, you will not need to re-spin the PCB. You can simply do a firmware update and boom, new product.

Each of the topologies you've mentioned are VERY refined, especially V.90 and ethernet and IR. Each has had its design absoultely squeezed to reduce cost over the years. each one has it's own ASIC (in this case, basically a program + processor + RAM all in one chip.)

If you were to BUY *GASP* the IP core *GASP* for a good processor, buy the IP core for a v.90 modem, buy the IP core for ethernet nics, and implement the hardware / firmware for IrDA yourself (its a very easy protocol and the hardware COULD NOT BE SIMPLER, you could produce this product yourself with all LICENSED IP CORES and be "in production" in your basement in 1 month.

Your biggest cost would be licensing of existing, very refined IP Cores, the software tools, and the FCC listing (See Part 15.239 and Part 62 (I think that one is POTs lines connected gear [v.90 modems].)


What you have to weigh are the usual constraints:
Either you want to spend NO MONEY and do it all yourself, all Core development, and PCB assembly (where the only cost is the time you put in and the cost of parts used.) This takes THE LONGEST TO GET TO MARKET. Leads to greatest $$$ per unit of sale.

or you want to spend LOTS OF MONEY, buy all the needed IP, tools, and get to market in a few months. Leads to lower $$$ per unit of sale.

or you can pay someone offshore to do it for you (no support). Leads to mid-level $$$ per sale, but will probably ruin your company in the long run when changes are needed if your developer disappears.