Article: 54754 of ba.internet From: Jeff Liebermann Newsgroups: ba.internet Subject: Re: Metricom 2.4Ghz interference? Date: Mon, 21 Feb 2000 23:25:35 -0800 Organization: Committee To Maintain an Independent Xenix Lines: 99 Message-ID: References: <38B2F779.F3714E7B@rage.net> Reply-To: jeffl@comix.santa-cruz.ca.us X-Complaints-To: newsabuse@supernews.com X-Newsreader: Forte Agent 1.7/32.534 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Path: news1.meer.net!news3.best.com!news2.best.com!newsfeed.berkeley.edu!remarQ70!remarQ.com!supernews.com!rQ66!corp.supernews.com!not-for-mail Xref: news1.meer.net ba.internet:54754 On Tue, 22 Feb 2000 12:54:17 -0800, Greg Retkowski wrote: > We've been using a 2.4Ghz wireless system to connect our house in the >santa cruz mountains to a place in Santa Clara where we've got ADSL >taking it from there (no good low price connectivity up in them hills!). >Late last year we noticed a degradation of our normally peppy 1-megabit >link down to the valley. It's slowly gotten worse. Right now its stepped >down to 384K, but can barely manage to pass 28-56K; latency has worsened >too. Rules of the game: 1. The first step to solving a problem is to assign blame. 2. The surest sign of success is pollution. 3. One measurement is worth about 100 guesses. Is the make and model of your hardware a secret? Most 2.4GHz point to point hardware will do 10Mbits/sec. If all you're getting is 1Mbit/sec when it was working, there's something fishy. Traceroute shows that your wireless ISP is hmmwv.net. The web page claims they have customers up to 13 miles away. How far away are you? You may be getting limited performance simply because the other end at HMMWV.NET is over subscribed. That basic bandwidth per radio is 10mbits/sec. Stick 10 customers on the system, and each will get 1Mbit/sec. 100 customers will average 100Kbits/sec (actually much less due to collisions). How many customers does HMMWV have on each of their 3ea radios? >The only event we can correlate to this performance degradation is the >rollout of the new Metricom / Ricochet 128K network, which is based on >2.4Ghz technology. Repeated requests to metricom for their frequencies >in this band (so we can re-tune to avoid) have gone unanswered. The reason Metricom isn't answering is that it's a non-question. They use the same 2.4GHz ISM band as everyone else that pushes wireless data. There's nothing to retune. Metricom poletops use three frequency bands. The 900MHz ISM band for the older RF modems, the 2.4GHz unlicensed ISM band for the news RF modems, and the new 2.3GHz frequencies for poletop to poletop. Since your unspecified hardware is in the 2.4GHz band, then you should not have any problems with the 915-928 or 2.3GHz transmitters. I won't discuss the current debate over 802.11b (11Mbits/sec) inteference with older systems. The 2.4GHz band is shared with microwave ovens, vehicle location systems, burlar alarms, vehicle location systems, various ISP's, video TV extensions, wireless LANs, MMDS trashmitters, and "wireless cable" video. If your unspecified location in the Santa Cruz mountains views much of metropolitan Santa Clara, then it's highly likely that you're looking at one or more of the aformentioned sources of potential interference. Your unspecified 2.4GHz radio probably has an RSSI (receive signal strength indicator) used for aligning the antenna. Most of these can be set to scan for sequences and therby be used for measuring interference. Find the RSSI output (it may be accessible via SNMP) and wave the antenna around looking for the source of interference. A 1000 watt microwave oven leaks far more crud than a city full of 1 watt Metricom transmitters. Your neighbors wireless VCR might also be the culprit. My guess is that there is another ISP, using the 2.4GHz ISM band, inline with your unspecified path to HMMWV. Check if your path crosses downtown San Jose, where the roof at AboveNet is plastered with 2.4GHz radios. A user that accidentally lands on the same spread spectrum code sequence as yours will fatally trash connectivity. Have you tried a different code? If this has no effect (or you have no control), perhaps diving into your unspecified radios SNMP or diagnostics will yield a bit error rate value. Interference does not "clobber" the signal. What it does it increase the base line noise level until the error rate creeps up. My guess is that this is what's happening. Santa Clara to the mountains is a fairly long radio path. Lacking sufficient information, my map shows a path of no less than 3 miles with a possible path as long as 10 miles. This is a big stretch for ISM radios. You need big antennas, 1 watt transmitters, and a concerted effort to avoid viewing co-channel users. Since you're getting only 1Mbit/sec, my guess is that you started with a marginal path, and that some source of interference is making it worse. Increasing the antenna gain may help, but if the interference source is inline with your path, more gain will simply amplify BOTH the desired signal and the interference. >Is anyone else in the Santa Clara area experiencing similar problems >with 2.4Ghz equipment? If you are, or have some info on the frequencies >metricom is using let me know. Thanks! Every one of my RF links has had interference or path imparment problems at one time or other. However, I have none on the Silly Clone Valley side of the hill. Generally, moving the antennas around, or sniffing for the culprit has eliminated the problem. In one case, I removed the radio and moved it very close to the other end of the link. It still didn't work so I knew it wasn't interference or path imparement. I had a broken radio instead. -- Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060 (831)421-6491 pgr (831)426-1240 fax (831)336-2558 home http://www.cruzio.com/~jeffl WB6SSY jeffl@comix.santa-cruz.ca.us jeffl@cruzio.com Article: 54984 of ba.internet From: Jeff Liebermann Newsgroups: ba.internet Subject: Re: Metricom 2.4Ghz interference? Date: Sat, 26 Feb 2000 01:49:28 -0800 Organization: Committee to Maintain and Independent Xenix Lines: 105 Message-ID: <8b8fbs0r2t0br49fcke0dgr405kiepdv38@4ax.com> References: <38B2F779.F3714E7B@rage.net> Reply-To: jeffl@comix.santa-cruz.ca.us NNTP-Posting-Host: p-088.newsdawg.com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Newsreader: Forte Agent 1.7/32.534 Path: news1.meer.net!news3.best.com!news2.best.com!newsfeed.berkeley.edu!pln-w!spln!extra.newsguy.com!newsp.newsguy.com!enews4 Xref: news1.meer.net ba.internet:54984 On Thu, 24 Feb 2000 17:14:38 -0800, Greg Retkowski wrote: [Reposted via Newsguy after Supernews ate this article.] >Its a pair of Arlan 655's which out of the box did a max datarate of 2Mb. >They are stepped down for reasons you'll see below. For more information >than anyone would want to know about our wireless equipment see >http://www.rage.net/wireless/. Nice HOWTO. Some problems. 1. 10ft of RG58c/u is far too long. At 2.4GHz, the foam dielectric flavour has a loss of about 3.2db or half your power is lost each way. There's also a good chance that it's full of water which makes things worse. I have some pressurized boxes you can mount at the antenna to cut your losses. 2. Your location is horrible. You're apparently shooting through a canyon and are certainly suffering from knife edge and Fresnel zone diffraction. The latter is where you have two signal paths between endpoints. One is direct, the other is bounced off the ground or nearby building. If the path length difference is a multiple of 1/2 wavelength, you get cancellation. I consider it a major miracle that it even works. I have a path profile program (http://www.topo.com) that will show the exact path but I don't know where hmmwv.net is located. Dig through: http://hydra.carleton.ca/info/wlan.html in the links and articles section and see the article on antennas and path characteristics including diffraction phenomenon: http://hydra.carleton.ca/articles/ve3jf-dcc97.html Also, you might find my mess at: http://www.cruzio.com/~jeffl/nooze/wireless.htm of interest. 3. The FCC *REQUIRES* manufacturers to use weird connectors so that people like you and me won't connect high gain antennas to their radios and do what you, me, and everyone else is doing. Just look at the weird antenna connector on the Ricochet modems. It's intentional. 4. The games problem is caused by the inherent latency introduced by the wireless link being half-duplex and using fast turn-around to simulate a full duplex link. Other than switching to a real full duplex link, it's not going to go away. I scribbled a news posting on this at one time and can probably find it if y'er interested. >That 'customer' is us. Reality is that we've got a partnership with this >guy whereas he's got a rooftop we can see and ADSL service; we (in the >mountains) put up the wireless system and everybody splits the ADSL bill. >We are the only network/wireless antenna on his roof. Nifty. A competitor. I do the same thing to areas not serviced by PacHell DSL. However, I use more expensive and better radios. >I finally got in touch with someone there; they are using freq-hopping >throughout the whole 2.4Ghz band; although they also said they have very >little traffic passing on their test network so it should not be a >problem. Ummm, not quite. There's very little *USER* traffic on their 2.4GHz poletops. However, when they switched to dual band (900/2.4) poletops, all the backhaul traffic (poletop to poletop, and poletop to WAP) went to 2.4Ghz, which made the 900MHz system much faster. >Unfortunately the Linux driver for our cards does not support access to >this data, and because of the age of the cards the developers of the >driver have moved on to the newer Arlan units. Ummm... Every wireless card I've ever seen has some kind of signal quality, level, error rate, alarm, or whatever available somehow. It's sometimes a test point, external jack, SNMP accessible, or just a register to poll. Using LRP rather limits what can be done. I use: http://www.freesco.org There's also: http://www.mikrotik.com which has traffic management features and a JAVA console that look useful. I'll do some reading and see what your cards can do. >At this point I'm assuming 2.4Ghz is falling victim to its own popularity. >We've played with different frequencies and datarates and gotten some of >our performance back but its still not where it was before... My revised guess is that the Santa Clara end is getting trashed by some kind of interference. I can drag a borrowed spectrum analyzer up to your place and see what it hears. Probably nothing unusual. However, the other end is in the middle of RF hell and will probably show considerable RF pollution. Add a bit of Fresnel zone diffraction and you're path is trashed. >My roomate is working with some of the newer cards, which have higher >power outputs.. I suppose if you can't carry on a conversation in a >crowded room you just scream louder. Ah well, time to move on to the >5.7Ghz band. Ummm, I don't think that 5.7Ghz is the answer. Reasons when the phone stops ringing... -- Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060 (831)421-6491 pgr (831)426-1240 fax (831)336-2558 home http://www.cruzio.com/~jeffl WB6SSY jeffl@comix.santa-cruz.ca.us jeffl@cruzio.com Article: 55010 of ba.internet From: Jeff Liebermann Newsgroups: ba.internet Subject: Re: Metricom 2.4Ghz interference? Date: Sat, 26 Feb 2000 18:10:13 -0800 Organization: Committee To Maintain an Independent Xenix Lines: 21 Message-ID: References: <38B2F779.F3714E7B@rage.net> <8b8fbs0r2t0br49fcke0dgr405kiepdv38@4ax.com> Reply-To: jeffl@comix.santa-cruz.ca.us X-Complaints-To: newsabuse@supernews.com X-Newsreader: Forte Agent 1.7/32.534 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Path: news1.meer.net!news3.best.com!news2.best.com!news.maxwell.syr.edu!europa.netcrusader.net!209.249.97.47!remarQ-easT!remarQ.com!supernews.com!rQ66!corp.supernews.com!not-for-mail Xref: news1.meer.net ba.internet:55010 On Sat, 26 Feb 2000 01:49:28 -0800, Jeff Liebermann wrote: (blah-blah-blah) Ah, I see the problem. You're using vertical polarization on your link. A bad idea. Metricom uses vertical polarization on all their poletops and handhelds. Since you're only interested in connecting to one other radio, switching to horizontal polarization will be a *BIG* improvement in interference reduction. Obviously, you have to rotate the antennas to horizontal polarization on both ends of your link. It appears that http://www.rage.net/wireless shows your antenna as vertically polarized. -- Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060 (831)421-6491 pgr (831)426-1240 fax (831)336-2558 home http://www.cruzio.com/~jeffl WB6SSY jeffl@comix.santa-cruz.ca.us jeffl@cruzio.com Article: 55014 of ba.internet From: DarkFiber@hotmail.com (Dark Fiber) Newsgroups: ba.internet Subject: Re: Metricom 2.4Ghz interference? Date: Sun, 27 Feb 2000 04:30:08 GMT Organization: Light Me Lines: 20 Message-ID: <89a9e8017fp@enews4.newsguy.com> References: <38B2F779.F3714E7B@rage.net> <8b8fbs0r2t0br49fcke0dgr405kiepdv38@4ax.com> Reply-To: darkfiber@hotmail.com NNTP-Posting-Host: p-046.newsdawg.com X-Newsreader: Forte Free Agent 1.0.82 Path: news1.meer.net!news3.best.com!news2.best.com!news.maxwell.syr.edu!newsfeed.cwix.com!pln-w!spln!extra.newsguy.com!newsp.newsguy.com!enews4 Xref: news1.meer.net ba.internet:55014 Jeff Liebermann wrote: >My revised guess is that the Santa Clara end is getting trashed by >some kind of interference. I can drag a borrowed spectrum analyzer up >to your place and see what it hears. Probably nothing unusual. >However, the other end is in the middle of RF hell and will probably >show considerable RF pollution. Add a bit of Fresnel zone diffraction >and you're path is trashed. Not to be picky, but the noise floor up on a hill (as seen through a spectrum analyzer) is MUCH higher than it would be on the valley floor. Last time I drug my Motorola 8920 up to Monument, the noise floor was incredible and without good notching, even with spread spectrum, you were screwed. I think you have your ends reversed. Article: 55016 of ba.internet From: Jeff Liebermann Newsgroups: ba.internet Subject: Re: Metricom 2.4Ghz interference? Date: Sat, 26 Feb 2000 23:36:21 -0800 Organization: Committee To Maintain an Independent Xenix Lines: 53 Message-ID: References: <38B2F779.F3714E7B@rage.net> <8b8fbs0r2t0br49fcke0dgr405kiepdv38@4ax.com> <89a9e8017fp@enews4.newsguy.com> Reply-To: jeffl@comix.santa-cruz.ca.us X-Complaints-To: newsabuse@supernews.com X-Newsreader: Forte Agent 1.7/32.534 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Path: news1.meer.net!news3.best.com!news2.best.com!news.maxwell.syr.edu!europa.netcrusader.net!209.249.97.47!remarQ-easT!remarQ.com!supernews.com!rQ66!corp.supernews.com!not-for-mail Xref: news1.meer.net ba.internet:55016 On Sun, 27 Feb 2000 04:30:08 GMT, DarkFiber@hotmail.com (Dark Fiber) wrote: >Jeff Liebermann wrote: > >>My revised guess is that the Santa Clara end is getting trashed by >>some kind of interference. I can drag a borrowed spectrum analyzer up >>to your place and see what it hears. Probably nothing unusual. >>However, the other end is in the middle of RF hell and will probably >>show considerable RF pollution. Add a bit of Fresnel zone diffraction >>and you're path is trashed. > >Not to be picky, but the noise floor up on a hill (as seen through a >spectrum analyzer) is MUCH higher than it would be on the valley >floor. > >Last time I drug my Motorola 8920 up to Monument, the noise floor was >incredible and without good notching, even with spread spectrum, you >were screwed. > >I think you have your ends reversed. Nope. The Santa Clara end is on top of some office building with a great view of all the 2.4Ghz Metricom junk deployed in the San Jose area, where Metricom is going to be pushing ISDN speed wireless. The Santa Clara end also has a good view of the nearby office building with their leaky microwave ovens, as well as going over residential areas infested with TV "rabbit" extensions, cordless phones, and whatever. Much of this stuff is fairly close to the Santa Clara end, which really what is important when considering interference (perverse square law and all that). Meanwhile, the other end is up above Lexington Reversoir, just below Skyline, at least 4 miles away from civilization and the potential sources of RF pollution. He's also shooting through a relatively narrow "valley" which limits what can be viewed. Therefore, my (revised) guess is that the problems is on the Santa Clara end. The reason Monument Pk in Fremont is so polluted is that Monument is one of the two transmitter locations (the other is San Bruno Mtn) for WavePath (http://www.wavepath.com) which provides MMDS data using 2.1 to 2.6Ghz. I'm not sure how much power they're belching, but it's quite a bit. With 500MHz to play with, they try to avoid the ISM band, but they manage to spray RF in there anyway. Also, mixed in between the MMDS stuff are some video links. Nice mess. In any case, methinks switching from vertical to horizontal polarization will make a BIG difference in avoiding Metricom interference. -- Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060 (831)421-6491 pgr (831)426-1240 fax (831)336-2558 home http://www.cruzio.com/~jeffl WB6SSY jeffl@comix.santa-cruz.ca.us jeffl@cruzio.com