Calculating 11a PtP Backhaul Requirements

If you are like me, and you are more familiar with the routing/switching/systems side of networking, calculating bandwidth over wi-fi can be tricky. In the hopes of saving others who were in my position a lot of time, I’ve sketched out some steps I took in calculating the Mbps requirements for several clusters of public safety camera in the Northern CA area using the unregulated U-NII and ISM bands used by 802.11a/b/g/n wi-fi. (This is very new, and I’m sure the numbers will change).In my example, let’s assume I have 8 clusters, each of which as 4 access points and 4 cameras. The cameras plug into the LIM ethernet port of the access point and then transmit via 4.9 GHz and PtMP (layer 2) to the central “hub” of the cluster, containing two radios — one dedicated to the PtMP connections, and another one dedicated to backhauling the traffic to another radio 5 mi away that is (supposedly) within LOS (line of site).

Let’s now assume we want to possibly build out the clusters, which might increase the backhaul requirements. In this case, we might need to change from backhauling via 4.9 GHz (50 available MHz, 4940-4950 MHz) and using one of less crowded channels on the U-NII bands (lower: 5150-5250 MHz; middle: 5250-5350/5470-5725; upper: 5725-5825). For the sake of argument, we’ll pick the U-NII band, which as you probably know, is the same frequency currently used by 11a (and with a firmware upgrade in the near future, 11n, which will give us 3 non-overlapping channels on 2.4 GHz and 8 non-overlapping channels on 5 GHz, and a typical data rate of 74Mbps thanks to MIMO).

Before I propose a solution, I need to make sure a few things: (a) that wifi can make that distance, (b) that I can make that distance using appropriate EIRP (”power”) levels, and (c) I can push the right amount of bandwidth through that long distance (and if not, how many intermediary links do I need). Roughly speaking, here are some steps to consider:

1. Get the best directional antenna at your disposal, in my case, it’s a 23 dbi panel antenna. (These types of antennas often shoot their signal in a polarized form, and you can use simultaneous channels by simply turning the panel 90 degrees)

2. Find how much power you’re able to use legally with that antenna. There are two situations — PtP (point to point) and PtMP (point to multi point), each of which has differing requirements for 2.4 and 5 GHz. Skipping over the complexities and nuances of this regulation, we’re limited to 52 dbm (~160W) on the upper 5725-5825 MHz U-NII band) for a PtP link.

29 dbm intentional radiator power (in radio) + 23 dbi panel antenna gain - 2 dB for cable loss = 49 dbm, just a tad bit under the 52 dbm FCC limit

(Note: if you do not know the difference between dB, dbm, and dbi, check out this resource)

3. Next, we calculate free space optic path loss in upper U-NII 5.725-5.825 GHz part of the 802.11a band. This is the amount of dbi we lose due to the laws of physics (i.e. wifi being wifi, nothing to do with anything obstructing its path). It is dependent on distance (in mi) and bandwidth (in Mhz)

20 * Log10 (5725 MHz) + 20 * Log10 (5 mi) + 36.6 = 125.7 dbi

(Here is an online calculator, and here is how that formula is derived)

4. Assuming your Fresnel Zone is cool (i.e. nothing blocking the greater area between you and the link), you now need to calculate RSSI signal strength that reaches the other point that is 5 mi away.

The general forumula for this is…

tx power - cable loss + antenna gain - free space path loss + rx antenna gain =rssi.

assuming a 2 dB loss in cable resistance, we can now plug and chug…

29 dbm - 2 dB + 23 dbi flat panel - 125.7 dbi + 23 dbi flat panel on other end = -52.7 dbm

5. I know look at the sensitivity rating of radio I’m going to use. In this case, our BelAir radios give the following mudulation rates at various RSSI levels.

54 Mbps: -71.4 dbm
48 Mbps: -73.4 dbm
36 Mbps: -78.4 dbm
24 Mbps: -81.4 dbm
18 Mbps: -84.4 dbm
12 Mbps: -86.4 dbm
9 Mbps: -88.4 dbm
6 Mbps: -88.9 dbm

6. Look at percentage lost because of distance. The general rule of thumb here is the 6 dB rule (3 dB is 2x, so -6 dB is -4x), where every doubling of distance results in a loss of 6 dB (this breaks apart a little when you compare 2.4 GHz vs 5 GHz, but it’s still more true than not true). I have a little cheat sheet, which shows me percentage I lose in terms of km.

I convert for miles (5 mi * 1.6 km/mi = 8 km) and then look on my chart to find the 10km mark. At that distance, I only have 52% of the 54 Mbps modulation rate, from step 5. “Good put” (what the radio actually puts through) is about 24 Mbps, so 52% of my goodput is, approximately, 12 Mbps.

7. Steps 1 through 6 assume we have an 11a radio, which is only sort of true. The BelAir ERM radio uses the atheros chipset, which gives us a few 11n-ish features, such as packet concatenation. The “good put” on these radios is around 35 Mbps, so dividing that in half leaves us a little wiggle room for the anticipated 16 Mbps (4 x 4 Mbps) that each cluster will have to backhaul 5 mi away using our 23 dbi antenna.