The following is a transcript of a recent email conversation I had with Sven Lid\u00c3\u00a9n, one of the primary network topology designers for the Sprint wireless network in the state of Washington.<\/p>\n
\r\n-----BEGIN PGP SIGNED MESSAGE-----\r\nHash: SHA1\r\n\r\n> -----Original Message----- From: C.J. Adams-Collier\r\n> [mailto:cjcollier@colliertech.org] Sent: Wednesday, February 08,\r\n> 2006 5:48 PM To: Sven Liden Cc: Andrew Weller Subject: Everett\r\n> Wireless Network discussion\r\n>\r\n> Heya Sven,\r\n>\r\n> Can you estimate for me how many radios would be required to cover\r\n> the non-metro area of the city of Everett and County of Snohomish,\r\n> if they were placed on utility (PUD) poles?\r\n>\r\n> Would the number differ if the radios were placed on cell towers?\r\n>\r\n> http:\/\/colliertech.org\/~cjcollier\/images\/maps\/\r\n>\r\n> Thanks!\r\n>\r\n> C.J.\r\n>\r\n>\r\n> -- <cjcollier@colliertech.org>\r\n> http:\/\/cjcollier.livejournal.com\/tag\/ +1 206 226 5809\r\n\r\n\r\nSven Liden wrote:\r\n\r\n> Pretty loaded question...\r\n>\r\n> It depends on a lot of things -- how high the antennas are above\r\n> the \"clutter\", what's their EIRP (power output) and what type of\r\n> antenna (gain) is used, and what capacity you want out of each one.\r\n> Plus, how much building penetration do you need (in people's\r\n> houses? In office buildings or store-fronts?)\r\n>\r\n> There are all kinds of ways to estimate it with software (I've been\r\n> out of wireless for 6+ years, so I don't know what the lastest\r\n> is), but even the software is limited by the granularity of the\r\n> \"clutter\" data you can get, which is expensive to buy.\r\n>\r\n> Generally, systems like this are modeled using a slope-intercept\r\n> model for each antenna. The slope is an estimate of the far-field\r\n> pathloss.\r\n>\r\n> You would typically do a \"link budget\", which calculates the entire\r\n> system loss... start off with the minimum detection threshold at\r\n> the receiver (the sensitivity), subtract the fade-margin, receiver\r\n> interference margin, and you end up with a minimum signal level for\r\n> on-street coverage. Then you subtract out pathloss for different\r\n> types of building penetration (maybe 10db for suburban, 15db for\r\n> urban, 20 for dense urban), and work your way back to the antenna,\r\n> and you can estimate how far the signal will go in free-space,\r\n> given the amplifier output and antenna gain.\r\n>\r\n> For free-space pathloss, you can just use a calculator like this:\r\n> http:\/\/www.distributed-wireless.com\/calculators\/pathloss_RSSI.html\r\n>\r\n> The issue is that pathloss in non-urban environments (like the\r\n> trees around Everett) can be pretty high. There are 80' trees in\r\n> some of those areas! Most wireless companies do a lot of\r\n> drive-testing to verify their models. We did this extensively in\r\n> Seattle because of the difficult terrain and dense foliage.\r\n>\r\n> I would guess that if the utility pole were only 35-40' it's below\r\n> tree level in most areas, and it won't carry more than a few\r\n> hundred yards, maybe 1\/4 mile... even with a high-gain antenna and\r\n> powerful amp.\r\n>\r\n> I could give you an \"order-of-magnitude\" guess, but that's about\r\n> it. You really need to sit down for a few hours and do the\r\n> calculations, look at maps, etc. For reference, the cellular\r\n> companies have about 300-500 sites each to cover the I-5 corridor\r\n> and metro areas.\r\n\r\n\r\n-----BEGIN PGP SIGNATURE-----\r\nVersion: GnuPG v1.4.1 (GNU\/Linux)\r\nComment: Using GnuPG with Mozilla - http:\/\/enigmail.mozdev.org\r\n\r\niD8DBQFD9QK1bS8rWWzCfqgRAnQiAJ4uDb9g5ZEwvcNWyKBMfl\/kv9znLgCgpMQW\r\njFe+nQ4vc6KhT+xcKnNZnWQ=\r\n=rQc7\r\n-----END PGP SIGNATURE-----\r\n<\/pre>\n\n