Hacker News new | past | comments | ask | show | jobs | submit login
Per Vices (YC W12) Sets Out To Disrupt Wireless Communications (techcrunch.com)
99 points by dwynings on April 30, 2012 | hide | past | favorite | 22 comments



If anyone is interested, I built a vaguely similar thing a few years ago [1]. The major difference is that it's restricted to the 2.4GHz bands, though the transceivers are modular and can be swapped out for broadband units. It's also MIMO capable, accepting up to 6 transceivers on a single PCI card. Being a little dated, the interface is PCI and size/speed of the FPGA is limited. Software is rudimentary, my ambition being to one day port GNUradio onto it. I've always intended to continue the development, but, so far, other things have intervened.

Full design details are released under the GPL and contained in the downloadable thesis [2].

[1] http://home.exetel.com.au/dalton/testbed/

[2] http://home.exetel.com.au/dalton/thesis_jd_toplevel_2009_11_...


I wouldn't be surprised if they offer you a job. You sound like the ideal first engineering hire.


Software Defined Radio for $11

http://www.reddit.com/r/RTLSDR

http://hardware.slashdot.org/story/12/03/31/1914217/software...

"DVB-T sticks based on the Realtek RTL2832U can be used as a cheap SDR, since the chip allows transferring the raw I/Q samples to the host, which is officially used for DAB/DAB+/FM demodulation. The possibility of this has been discovered by the V4L/DVB kernel developer Antti Palosaari."


Their site is a little optimistic for my taste. Awesome company name, though.

"Phi can capture over the air waves, so with the right app, you can watch cable for free."

Cable != over the air.

"Phi will be able to act as a base station to let you make free calls using your cell phone."

AFAIK this is illegal in the US and even if it was legal the carriers would try pretty hard to block it. I'm up for a good disruption as much as the next HNer, but this sounds more like a long-term aspiration than a product feature.


Yeah, I was under the impression that it would be illegal also, but apparently the FCC hasn't concluded whether it should be allowed.

MagicJack was planning to offer similar functionality in the past:

"YMAX [makers of MagicJack] says it doesn’t need permission to use the carriers’ operating frequencies in the home and that it’s protected under Part 15 of FCC regulation. This stipulation allows a device to operate without a license if it uses very low power, and doesn’t cause interference."

"In 2007, the CTIA requested the FCC regulate in-home wireless network signal boosters. The CTIA believes it should be illegal to boost any signal without the consent of the spectrum licensee. This contentious issue is still being debated, and YMAX is likely watching very closely for any signs of a decision."

http://www.audioholics.com/news/industry-news/magicjack-cell...


But if it works in other markets, it could make it back here. I think a lot of people discount the possibility of introducing disruptive technology in major international markets that are also developing economies. The BRICs plus Africa would be interesting markets for disruption in the telecommunications infrastructure space.


I'm wondering if you have any tips for people who want to apply to YC with a non-software idea? Or some info as to what they can expect once they get there? It's got to be a different experience.


That's an interesting question. Certainly, a lot of the advice is aimed at software companies - and it's sometimes hard to figure out where to draw the line...

For example, the idea of a minimum viable product doesn't really work too well with Hardware. Neither does shipping fast and iterating - hardware recalls are expensive.

For us, the best value in YC was the partners. Having smart people to go to when you're facing problems is probably the most valuable part of YC. A lot of times, your problems have very little to do with your technology or industry.


"For example, the idea of a minimum viable product doesn't really work too well with Hardware."

Actually, it does, if you make your MVP a little more minimal. Put up a sales page for the product, attract traffic with Google AdWords, and see who clicks "buy".

You know those infomercial products that required 6-8 weeks for delivery? That was the time it took to actually get the products manufactured, which didn't happen until a sufficient quantity of orders had been reached. Tim Ferris has a bunch of similar examples in 4HWW.


Congrats Guys. Per Vices has all the signs of a product that, like Apple, starts out as a hobbyist project for hackers and ends up disrupting the world. The Internet has disrupted TV channels and movie studios, and now the wireless monopoly is the last bastion that these guys will take down.


1. Can the PCI card transmit in any frequency (up to 4GHz)? Is this allowed by the FCC?

2. Wireless protocols are timing sensitive. Can the card load up a program to process signals? Similar to a kernel in a GPU?


Disclaimer: I have no more knowledge than what's up on their page, a lot of this is professional guesswork.

1. It most likely can. They are probably hoping to not bother the FCC by treating it as "test equipment" (that's what Ettus Research does), leaving the responsibility of compliance with the end-user.

1b. I doubt they're getting as much as a Watt out of that board (personal estimate would be a peak of 10-50mW, [10-17 dBm]) so in ideal conditions, significantly less power than a cell phone or wifi signal. Still enough to be trouble in some bands, and it doesn't change the regulatory situation.

2. There is an FPGA, so there is a physical possibility. It's not a trivial thing though, and working with FPGAs is even more remote from what most programmers are used to than working with GPUs. If I had to guess, the plan would be to capture/transmit bursts, with processing in between, instead of real-time signal/protocol shifting.


Dear Patrick,

I'm one of the founders of Per Vices. To confirm;

1. We are currently selling our device as test and measurement equipment. The intended use case is academic or research.

2. You're mostly correct about power - although it turns out that, depending on regulatory jurisdiction and the frequency or band of interest, power level can matter.

3. Actually, we're aiming for real time programming using gnuradio. You're absolutely right about the bursting, though - our initial programs were almost exactly as you described.

Long term, one of our driving goals is to drastically reduce the barriers (specifically cost and expertise) standing between the average software programmer, and a 'hello world' program that can usefully transmit information.


Thanks for weighing in. I hope I don't come off too aggressively negative. I used to work for thinkRF (nearby, in Ottawa), and I think I was just exposed to too many people who counted on software to magic away everything. I feel like I need to keep enthusiasm in check, having seen the problems people will run in to.


That's perfectly valid - I don't immediately disagree with that approach; especially because there are many things software can't fix.

But I also think it's much easier to address problems in software than in hardware...


Does it look like Per Vices' product is similar to Ettus'? Maybe cost being the significant difference?

I am not familiar with the specifics of wifi protocols. Is the wifi protocol fit into the rx/tx burst+processing or RT signal/protocol?


Biggest differences between our product, Phi, and the USRP (ettus) are cost and bus. We use a x4 PCIe, providing much higher transfer bandwidth (8Gbps vs 1Gbps). Additionally, the RF front ends are fully integrated (lowers cost).

We're aiming to have the user manual up by the end of the week, which should provide much more information.


I guess (stressing 'guess') you could do wifi fine, if you can afford a lot of dropped packets. You would listen for some short while, and process that signal to look for packets sent your way. When you needed to send data back, you pre-process it one packet at a time, ('packets' at the wifi level, not necessarily TCP/IP) and when you hava packet you just fire it out the radio.

There's a fair bit of work going on behind the scenes there, and wifi is a pretty heavy protocol to work with like this. I would be very surprised if Per Vices doesn't move to a seperate wifi chip in a future version, if they stick to the 'universal bridge' product. For such a popular protocol, it would make much much more sense.


I see an Altera FPGA on the card. If that's user-programmable, this thing will be awesome. Hell, I'll buy one in a heartbeat.


Looks like a Cyclone IV (at least I hope it is, because I would feel really bad if they implemented PCIe on the Cyclone II). Of course, whether you can get anything out of it will depend on if they've put on a version with enough gates to do something interesting.

The pessimist in me says no, but OTOH it is really nice to see something hanging off the PCIe bus, instead of trying to move all the data to the computer over USB or a network card.


It is a cyclone IV.

More specifically it's an EP4CXG22. Minimum speed grade is C7, but I think we ended up getting some C6's in there (doesn't really change much).


Is it accessible to developers? Really want to hack with this stuff.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: