What is a good level and/or SNR for WiFi?

[Dave2002]

I have been using Netspot to survey the wireless environment around me. I have been trying to extend the range of one Internet connection using a wireless access point and a powerline link to that from a relatively distant router.

So far it seems to me that "good" equates roughly to a signal level of or greater than -50 dbm, with an SNR of around 35 dB.

Not so good equates to a signal level between -50 dBm and -70d dBm, with an SNR of around 20 dB.

Poor seems to be below -80 dBm with an SNR around 10 dB.

Getting consistently good results across several different rooms seems hard to achieve, to the extent where I wonder if more than 2 wireless access points are needed. Currently I have actually 3 running, but on different services. The two which are on the same ISP are on different levels - one is upstairs in the hope of giving good coverage up there, and also for rooms below. I am running those two on different wireless channels deliberately, and the survey suggests that I have so far managed to avoid the channels used by the neighbours. "Good" is easiest to obtain when in the same room as the router or access point.

[Gordon]

Can't help specifically on this occasion Dave, but WiFi is defined by the IEEE 802.11x standard [x can be any one of a number of letters of the alphabet] which you will find on Wiki here:

http://en.wikipedia.org/wiki/IEEE_802.11

They give ranges in metres/feet but no SNRs I'm afraid, but a bit more googling might get you somewhere. I doubt that the manual with your routers [Ha!] will say anything that useful!!

You will see that they operate at very high frequencies between 2.4 and 5 GHz and so penetration through domestic walls will be variable depending what the walls are made of. One bad building material is plasterboard with a thin metal film for heat reflection but is bad news for radio waves especially in this frequency range.

The use of OFDM [as used in DAB and in DTT but different in detail]and a technique called Multiple Input Multiple Output [MIMO] means use of multiple aerials [you will see them sticking up out of the router, typically 2 but can be more] that helps with some difficult propagation problems but not necessarily with flat loss. When you have an envronment in which the radio wave is reflected a lot and can therefore interfere with itself [multipath] then a simple SNR isn't necessarily a great help to predict performance. The SNR to expect varies with the order of the OFDM too as it does in DTT. Some routers are able to adapt dynamically to the propagation conditions and so may work but sluggishly. Broadly speaking, Mr Shannon's rule that SNR and bit rate are reciprocally related applies, but applies best when noise is the limit - typically a flat loss environment - rather than multipath.

Your PL links should also adhere to 802.11 - but best check. Sorry can't help any more, it's one of those suck and see situations.

[Dave2002]

Interesting that you should mention MIMO. I have wondered for some time how it works, and also how well it works. The wiki article - http://en.wikipedia.org/wiki/MIMO is interesting, though I think the mention of wired connectivity using MIMO is incorrect.

It is not really clear to me how MIMO works in practice, or how it might work in a dynamic mobile environment. For a fairly static environment it may be possible to reduce some problems of multipath transmission, and indeed perhaps exploit multipath where signals reinforce. MIMO won't really be MIMO unless both the transmitter and receiver units have multiple antennae - so a MIMO base station sending to an older receiver will, I think, operate in MISO mode - if I understand the terminology correctly, though surely in the reverse direction it will be SIMO.

Spatial multiplexing is an interesting idea, and also the use of OFDM in an adaptive system could optimise the "channels" for that communication method by optimising each one. I have no idea whether "simple" consumer devices are capable of doing that. [Of course many consumer devices are nowadays complex - even an FM radio is quite complex, and DAB sets, whatever their faults, do many things in order to function.]

The BT Home Hub 5 appears to be a MIMO device, but perhaps with different configurations for different frequency ranges and transmit/receive modes - according to some reviews.

I have been extending the range of our wireless coverage by using powerline adapters linked to wireless access points. So far I have only used regular APs, not MIMO units. The results are promising, and give much better coverage, but I'm still not getting good coverage in all the target areas.

One other aspect of MIMO which may work is improved ability to separate out different devices in the same room. I found years ago that orientation and relative positioning of different laptops can make a very significant difference to performance, and perhaps MIMO can help with that. [I conducted repeatable experiments to verify the orientation effects, which were sometimes very significant.] It's not also clear whether most modern laptops or even desktops use MIMO - and have multiple antennae built in.

So finally, I am confused as to whether MIMO really works in practice, given that it's uncertain whether many of the devices in the coverage area may not themselves be MIMO "aware". Maybe I should try a MIMO wireless access point if I buy yet another one to extend our coverage further, though it might be cheaper to just use a regular one, and in the end just as effective. I'm expecting 802.11ac devices eventually to become cheap(er), but that has not happened yet, so I might just go with one more reasonably simple wireless access point, rather than a more expensive MIMO multi-standard one.

[johnb]

Not really relevant but a couple of years ago I was using a laptop on WiFi ('g') in an area of the house where the signal strength (as shown on the laptop) was 50% and less and I found a very dramatic decrease in speed.

[Gordon]

MIMO certainly works but the conditions influence its performance. The multipath problem is closely related to wavelength so antenna separation matters. In a consumer device compactness is important and so MIMO type processing loses some of its value. At 2.4 GHz the wavelength is about 300,000,000/2,400,000,000 metres = 1/8 or about 120 cm which is rather large compared to a router!! However in a vehicle say the antennas could be separated enough to get better results.

The full benefit is obtained in MIMO, as apposed to SIMO or MISO, because an SFN with several suitably spaced transmitters and/or several suitably spaced antennas from one transmitter will increase considerably the confidence of getting service continuity at a particular place in a service area. It's the antenna spacing that matters in principle not the receivers'. A MISO device in a MIMO network will not fare so well as a "proper" one!! But if it is placed in a favourable location in the coverage field it might do well.

DVB-T2 as used in DTT now has a MISO facility [DTT doesn't actually invoke it because it doesn't use SFN] meaning that there could be several transmitters in an SFN [like DAB] but only 1 antenna/receiver so that there are several rays reaching that antenna and the plan is that not all will be conflicing all the time so that continuity of service in harsh conditions will be better assured. The conditions prevailing in WiFi and for portable radio reception are such that there is normally no direct dominant path for the radio wave, what is received is a whole series of dynamically changing secondary paths. This so called Rayleigh regime is hard to predict precisely and so empirical statistical models are applied.

OFDM is actually very good on its own in these environments because the bit stream is spread over many closely spaced carriers so that if a mutlipath event sweeps through the band not all the carriers are affected simultaneously. At the receiver the FEC will catch that event and mitigate it. But as in all things the devil is in the detail and the design of the network and the OFDM must take acount of severe multipath if that is expected. For broadcast the Rayleigh model is not normally used as a basis for planners but it is taken into account if you look at say the revised plans for DAB published by OfCom a couple of years ago.

Clearly the distances involved in a broadcast network and in WiFi are very different so one must scale the design. WiFi as commonly used deploys a frequency plan that allocates users separate carriers. A MIMO one would need to allow several WiFi devices [which operate bi-directionally of course, unlike broadcast] to use the same channel so that they can reinforce each other. IOW a WiFi device downstairs will help the one upstairs as a laptop moves about if both devices operate together. If the laptop had more than 1 antenna it wold get extra support too.