Real World Examples of Device RSSI Comparison

By gstefanick posted Apr 14, 2015 12:47 PM


Wireless clients are without question one of the leading problematic components of a wireless network. Many WiFi engineers agree 80% of issues on todays WiFi networks are client related. Wireless clients are much like humans. They all speak, hear, and act a bit differently from one NIC/Driver to another. While you might observe similar or identical behavior across a specific device type, but a simple driver update can change everything you know about how that device behaves. 


In this blog post I will share 5 devices in two real world scenarios specific to how devices interrupt signal (RSSI). 



Scenario #1 


A hospital has a large deployment of Ascom i75 handsets. They augment the Ascom deployment with Cisco 7925 devices because of cost and a resent Cisco Call Manger installation. However, the hospital shifts back to Ascom because of nurse call and other Ascom integration features. The hospital purchases the Ascom i62 handset which supports 802.11a/b/g/n. The i62 is the replacement phone for the i75. 


But there is a problem. They deploy the i62 handset to a number of departments. Users are reporting voice issues. Loss of speech, roaming problems and clicks. Nothing has changed but the handset. Previously these departments had both the i75 and 7925 working without issue.


Let the troubleshooting begin. Focus on the basics first, do I have signal? What is the phones perspective of the signal. It is quickly noticed the i62 is consistently 9 dBm lower than its counterparts attached to the same ap in the same location. In fact the i75 and 7925 are consistent in testing. Assuming we might have a bad i62 we pull 5 more out of the box. Same results. Next is the firmware, so we check and we’re on the latest firmware. A call to Ascom revealed the i62 radio has a lower receive sensitively when compared to the i75.



Scenario #2 Intermec CK71 and Proxim 8494 


A warehouse is deploying a new wireless network. Part of the design phase we enquire about the type of devices that will be used. We are told two types of scanners. One a CK71 and the other a truck mount computer. We ask for two of each to test. The study of the truck mount computer revealed very similar RSSI behavior when compared to our survey laptop with the Proxim 8494. 


The Intermec CK71 didnt far as well. In fact the CK71 was consistently -8 dBm lower than our survey gear. Based on these results we knew that we needed to factor in the -8 dBm difference into our design. 


ck71.proxim.card 3.png


RSSI is only one metric of many to use when designing and troubleshooting a wireless network. Knowing how the client interprets the signal is very important. We just seen how the Ascom i75 and Cisco 7925 could be voice grade at -67 cell edge while the Ascom i62 would show -76 dBm at this same location! YIKES! 

Also designing to the lowest common denominator is critical to your assessment. Over the years I’ve conducted dozens of device studies. Each go into my personal archive for future device comparison. 

So have you seen any clients in the wild with an aggressive or wildly different receive sensitivity when comparing devices?



1 view


May 05, 2015 05:41 PM

Lol. This also reminds me of the saying "The packets never lie". That is a good saying in wifi... then you learn, "The packets never lie but your wireless card and decoder might" hahaha.

Sent from my Verizon Wireless 4G LTE smartphone

May 05, 2015 05:16 PM

George and "Airhead chief" you guys are funny being philosophical here LOL.  I never thought wifi can be so deep!

May 05, 2015 04:13 PM

I find myself always saying ?Yeah But? ? What if? ?Depending on: reference point, duty cycle, climate, orientation, human bodies, client capabilities? ? And so on. It is so hard now to be confident in your numbers and test data when it seems like everything can be off just a little bit due to many things. The moment you realize you can not be "absolutely" certain is when you become a Wireless Expert hahaha For real, I believe every engineer passes that threshold at some point. It take a little believe.. Or magic? :-)

May 05, 2015 03:58 PM

LOL.. So true .. Think of a end user who is told voice needs to be at -67 and your device says -70 its out of scope .. when it could very well be in scope .. 

May 05, 2015 03:44 PM

As I learned more and more about how relative wireless is? I realized I was not able to be as certain in things as I use to and as a result I seemed like I then knew way less. lol

May 05, 2015 03:20 PM

The problem is that you need to measure and baseline to have a point of reference. For the common user and engineer this is RSSI. Good, bad or ugly it what we have to use. So many balls in the air as you know .. Thanks for the response! 

May 05, 2015 08:39 AM

Hi George,

Excellent post from the "field experience", although I don't hang out that much in the field like you, I want to add something about the RSSI thing.

Most of the time, the radio vendor or devices they do not calibrate the RSSI/signal(dBm) at the input of the radio + the antenna.  From my own experience there can be a significant difference up to 10dB from the real value and the value reported by the chipset/driver, it depends lots on the chipset and the LNA(low noise amplifier).

For more accurate measurement, I had to "calibrate" by offseting what ever is needed to match to real signal.

As you know in 2015 there is this ETSI adaptivity thing in the Euro regulation, to meet new ETSI we need to know the exact input signal to comply with regulation, basically if the radio sees a signal above an absolute level, it has to stop txing, FCC we do not have these rules.

Our chipset vendor basically told us we need to "calibrate" because they did not have such measurement saved during the manufacturing process, they do calibrate txpower but not rxpower, they promise they will do such in future process to easy people like me from calibration.




Apr 27, 2015 12:07 PM

When I discovered this I thought the same thing. However after chatting with two different folks at Ascom I was told that the i75 and Cisco 7925 uses the identical Texas Instruments chips. While the i62 is a new chip. I was also told the Ascom Myco uses the same chip as the i62. 





Apr 27, 2015 05:45 AM

Hi g,


I work in hospital in france as wireless expert, and we have the two model (i75 and i62)

the i62 is a dual-band handset, my question :

do you think is possible that Ascom would have reduce this RSSI on the 2,4 GHz to obtain an equivalent  RSSI on the 5GHz ? 


thanks again for this excellent post !

Apr 24, 2015 11:26 AM

Hey Guys,


1) Many radios show dBm as a receive value, while others may show RSSI. I tend to use dBm becuase it better describes what Im looking at. But both can be used.

2) You are on target. I only mention dBm becuase I wanted to share what each device was showing indicually. But yes, when you compare it should be dB. 



Apr 22, 2015 06:56 PM

Great post I've been doing a ton of similar testing recently with medical devices, voice devices and mobile devices with celluar.   I also remember a similar scenario years ago at a hospital in Michigan where the WLAN was designed to support the B2000 Vocera badge but they still had some users trying to use a B1000 and they would drop calls and tried to blame in on the wireless survey.

Apr 21, 2015 03:12 PM

On your second thought:

RSSI is a measurement of the power at receive radio. mW is the unit of power. Decibel is the ratio between the two values. (my thought process)


So, using dBm helps identify the unit we are using. Using just dB would be like stating a ratio without identifying what you are actually measuring. Like 10 dB = 10, 17 dB = 50; but 50 what!? dBu, dBa, dBc

Either way should make sense in our world, since we know mW is the unit....

On your second thought:

I agree. lol


Thanks for the thought provoking post to help keep me awake after lunch! I look forward to what George has to say.....

Apr 21, 2015 02:28 PM

@gstefanick I have a few questions with the above blog and follow up post. I go back and forth with the terminology in books and how people use them in real life and just want to set it straight on here for others to learn. Help me out :-).


I thought:

  • RSSI is not measured in actual dBm but in an arbitrary set of numerical units. Like posted here.
  • Also, when we talk about the difference between 2 measurements, I believe we should use dB instead of dBm? Mentioned here.

What do you think?

Thanks for the help man.

Apr 14, 2015 02:49 PM

Great post George. I have also seen cases cause difference in RSSI. I have a very slim case on my phone that drops the RSSI by 4 dbm. I have seen times where testing was conducted with a standard device, then afterward cases were given to the users to help protect the device. Thankfully the design was hefty enough to not cause issues, but when interviewing clients this is an important piece of information to ask about.