Hacker News Reader

Comment by ghshephard | Hacker News Reader

"modern devices are down to less than a millisecond of latency per hop."

Excluding contention, and noise, which can result in missed channel transmit opportunities, latency on an RF Mesh is a function of the protocol (ALOHA, Slotted ALOHA etc...), packet size, and transmit speed.

In addition, if you are using the same radio transceiver for transmitting and receiving, you cut your performance exactly in half.

For a 100 kbit radio @ 256 byte packets, you are looking at approx. 50 msec/hop. It's not that surprising to have a 5 hop network in the field with about 400 msec latency before you get to the WAN backhaul.

The Tarana stuff looks interesting though - Here's hoping they can pull off everything they describe on their site.

View on HN

We most certainly can do what we say.

Check out http://youtu.be/o_3GUKsQ4CQ to see for yourself.

It's called "Seeing Is Believing" for good reason.

regards,

Steven Glapa (Tarana's marketing guy)

190m (623 feet) for this type of connection is not difficult. I was with a company where we were able to multipath in this fashion in 2001. This was with off-the-shelf radios and antennas in Part 15 rules (2.4Ghz & 5.8Ghz).

Making multipath work for you doesn't require smart radios/antennas.

Now, if you were doing 400mbit/s, using multipath, with dozens of radios, hundreds of clients across many square miles...that would be entirely different. That's a real world scenario. Show me a video of that and I'll be impressed.

I've been out of the wireless tech for a while, but the only companies I've heard make claims like this were never able to deliver.

Hi Steven, good to see myself (as a potential customer) in that video :)

I can attest that the performance were indeed those seen in the video. We are testing the system these days and, so far, it looks like nothing else we have seen.

Disclaimer: I'm the CTO of a large Wireless ISP.

Great that you're seeing the performance in your testing! Sorry about the unscheduled video appearance -- unfortunately I didn't have the time to apply the "blur" brush exactly everywhere. Let me know if you'd like that done and I'll take care of that for the final version :)

So NGI's Eolo is going to test it? (:

Ed -- Thanks for the healthy skepticism -- we get a lot of that. Yes, a single link in isolation at this range is relatively trivial, which is why we and our customers were interested in seeing what would break it. I doubt your no-smart-antennas 2001 example would have been as successful with one of its radios face down on the roof, pointing in the opposite direction, or blocked by an aluminum sign. We've seen that this kind of resilience makes backhaul network engineers very happy.

As for co-channel interference in a dense deployment, we've demonstrated that on our multi-site test network (with customers watching there, too) in Berkeley. That was before we had anyone on the team here to whom it had occurred that it might be useful to video the proceedings -- and since then we've been too madly busy on product work to worry about demo videos. Our capture from Italy a couple weeks ago was a bit of a bluebird.

I'll look forward to showing rather than telling as our deployments roll out over the next several months. Stay tuned, if you want to see the proof accumulate that not all wireless vendors lie.

Steven

looks like they are using multiple radios on the same frequency and then also are able to overcome self-interference. I read an IEEE article about receiving and transmitting on the same frequency at the same time by subtracting the transmitted signal from the received to get only the signal from the remote radio. from their website " Perfect cancellation of co-channel interference " They also mention 'beamforming'.

Reply on news.ycombinator.com