Wi-FiGurus

The community of Wi-Fi Professionals

SNR Cutoff Recommendations PDF Print E-mail
(6 votes)
Written by Jim Geier   
Monday, 25 June 2007
The performance of a RF site survey requires more than just ensuring that users are able to connect to an access point. Learn what signal-to-noise ratio (SNR) to use for defining effective range boundaries.  
When performing a RF site survey, it’s important to define the range boundary of an access point based on signal-to-noise (SNR) ratio, which is the signal level (in dBm) minus the noise level (in dBm). For example, a signal level of -53dBm measured near an access point and typical noise level of -90dBm yields a SNR of 37dB, a healthy value for wireless LANs. Don’t let the unit “dB” throw you – it merely represents a difference in two logarithmic values, such as dBm.

SNR Variance Impacts Performance

The SNR of an access point signal, measured at the user device, decreases as range to the user increases because the applicable free space loss between the user and the access point reduces signal level. An increase in RF interference from microwave ovens and cordless phones, which increases the noise level, also decreases SNR.

SNR directly impacts the performance of a wireless LAN connection. A higher SNR value means that the signal strength is stronger in relation to the noise levels, which allows higher data rates and fewer retransmissions – all of which offers better throughput. Of course the opposite is also true. A lower SNR requires wireless LAN devices to operate at lower data rates, which decreases throughput. 

Real-World Values

I recently ran user-oriented tests to determine the impacts of SNR values on the ability for a user to associate with an 802.11b/g access point and load a particular webpage. For various SNRs, here's what I found for the signal strength (found in the Windows radio status), association status, and performance when loading the http://wireless-nets.com/staff.htm webpage from a wireless laptop. To ensure accurate comparisons, I cleared the laptop’s cache before reloading the page:??> 40dB SNR = Excellent signal (5 bars); always associated; lightening fast.??25dB to 40dB SNR = Very good signal (3 - 4 bars); always associated; very fast.??15dB to 25dB SNR = Low signal (2 bars); always associated; usually fast.??10dB - 15dB SNR = very low signal (1 bar); mostly associated; mostly slow.??5dB to 10dB SNR = no signal; not associated; no go.

These values seem consistent with testing I’ve done in the past, as well as what some of the vendors publish. ?
SNR Recommendations
?Based on this testing, I recommend using around 20dB as the minimum SNR for defining the range boundary of each access point. That ensures a constant association with fairly good performance. Keep in mind that the corresponding level of performance only occurs at the boundary of each access point. Users associating with access points at closer range will have higher SNR and better performance.

When measuring SNRs, be sure to use the same radio card and antenna as the users will have if possible. A variance in antenna gain between the survey equipment and user device will likely result in users having different SNR (and performance) than what you measured during the survey. Changes made in the facility, such as the addition of walls and movement of large boxes, will affect SNR too. Thus, it’s generally a good idea to recheck the SNR from time-to-time, even after the network is operational.

Jim Geier provides independent consulting services and training to companies developing and deploying wireless network solutions. He is the author of the books Wireless LANs (Sams) and Wireless Networks - First Step (Cisco Press).



Jim Geier is an independent consultant and founder of Wireless-Nets, Ltd (www.wireless-nets.com), a consulting firm assisting municipalities, enterprises, hospitals, airports, and equipment providers with the development and deployment of wireless networks.
Comments
Add NewSearchRSS
outbackwifi (Registered) 2007-06-26 06:53:37

Dear Jim,
Excellent article. While I tend to agree with you, I have often found that most wifi cards (which use ndis drivers) do not report the noise levels at all, so the SNR is incorrectly reported as equal to the Signal. What do you suppose one could do in such cases?
chris_91504 (Registered) 2007-06-28 12:53:26

I guess you need to purchase site survey software from either Ekahau, AirMagnet or Helium that supports collection of noise values and calculates SNR in a relaible fashion. This would cost you a few hundred dollars in case of Helium or thousand or more with AM or Ekahau
Noise Not Correct - thus SNR N
Keith Parsons (IP:68.26.84.20) 2007-07-10 14:49:21

802.11 cards cannot 'see' ambient RF noise - only the 'effect' of noise on packets. Thus 802.11 cards are not very effective at all in reporting accurate 'noise' - only a SpecAn can do that.

I can easily show how to 'fake out' the noise functions of many cards - to either show lots of noise where there is little, or report no noise where there is a lot.

Yes - a nice gap from the reported noise floor to the reported signal strength is great - but just not reflective of actual 'RF Noise'
jimgeier (Administrator) 2007-07-11 13:21:52

You''ve all made good points here. I recommend using a spectrum analyzer (or other tool designed to accurately measure noise) to measure noise levels. You might consider what Cognio has to offer.
WiSpy: The Poor man's Spec An
outbackwifi (IP:59.176.64.45) 2007-07-16 12:24:39

I got my hands on the Wi-Spy this month. This is a tool from metageek. I will post a more detailed review later on. I could spot my 2.4Ghz cordless phone showing up along with my wifi signal.
Have any of you used it/come across it? What I like about it best is that it has a software that runs on Linux too (Though Kismet has support for wispy too)!!
karl1949 (Registered) 2007-10-08 16:41:59

I couldn't agree more with Jim. I do a lot of predictive designs with the Ekahau tool, and I ALWAYS use SNR as the main design criteria, along with resired data rate, network load, and the appropriate client card to determine my cell edges.
If the signal is strong and the noise is also strong, then the SNR is too low to support a good connection.
I explain it to my customers this way: Signal strength is a measure of the quantity of signal present, and SNR is a measure of the quality of signal present. Without good signal quality, it doesn't matter how much quantity you have.
As for the cost of the Ekahau and AirMagnet tools - totally worth every penny. I also carry a Wi-Spy with me as a "quick-and-dirty" means of demonstrating that interference is present, leading to a full spectrum analysis with Cognio - also a tool totally worth the price.
Write comment
Name:
Title:
UBBCode:
[b] [i] [u] [url] [quote] [code] [img] 
 
Security Image

Powered by JoomlaCommentCopyright (C) 2006 Frantisek Hliva. All rights reserved.Homepage: http://cavo.co.nr/

 
< Prev   Next >