Aka. Sometimes borrowing has to go a little far.

Literally here.

I gave a quick primer of how to borrow Wifi from a neighbor, but few places are as densely populated as NYC. Therefore in other residential neighbohoods, borrowing wifi is less trivial and may require some technical knowledge to know what challenges you are dealing with.

Respite from Obsessive Compulsive Disorder…

TLDR: A Panel antenna can get you excellent coverage, up to 50m. Most advertise wi-fi range at 300ft or 90m. Below is chart of distance vs data thru-put, on panel antenna on raspberry pi, to home router.

The groupings of points could be because of different AP I am measuring from. Different transmission power. In-accuracies in the distance measurement to the AP. However, in this case, I think is showing how the environment can change the transmission rate. This can be because of rain, radio interference, or obstruction in line of sight. This causes greater range of max bps, the further away the antennas are. However, you will notice that the centers of data at each distances, show loss of throughput.

Also we are assuming a linear relationship between signal strength and bandwidth. This is not correct, but a linear model will serve as a approximation, of our equipment and ordinary outdoor conditions.

But generally the data shows a loss of -378.3kbps per meter, though the data is far from clean. Significant loss of bandwidth is to be expected at 97m. And the linear model of performance says 36,194,630bps, is best possible performance, though there are measurements as high as 50Mbps. 36,194,630bps should be interpreted as maximum AVERAGE performance we can ever expect to see given the test scenario which includes interference and contention for airspace. Both 36,194,630bps and 50Mbps is far less than Wifi-n and it’s maximum of 450Mbps, even though both AP and receiver use Wifi-n. (the current standard today is wifi ax, as of 6/20/2024). This is probably bc the raspberry pi is so underpowered.

…Return to Destructive Detail Orientation

Logic behind throughput and Signal strength
There are a few simple variables most non-engineers can keep in their heads

1. Transmission power is limited to 26dBm in the USA, and in most countries. This is b/c Wifi is in a unlicensed radio band, labelled scientific and medical equipment, in USA. Anyone can build equipment for these frequencies, as long as they adhere to the limits in the regulations. Other licensed frequencies, they are bidded by, by large companies and comglomerations, to exclusive use.
2. All antenna equipment is limited by the minimum power the transceiver can receive, before it can understand it. This is analogous to a voice having to be at least a whisper before you can hear it. Noise matters, too. So having a lot of equipment in same range, they tend to interfere with each other. Not only does the receiver have a minimum power, but the received signal has to be slightly louder than the noise. On your computer, you have a Wifi transceiver and equipment, and often the drivers for the will allow access to the signal value. But not always. Sometimes you end up with a generic driver, that your wifi equipment has minimum compatibility and though it can associate, it won’t give you signal strength in dBm. Typically, these software drivers give wifi signal strength in a artificial number called RSSI. It is only relevant to the exact same model, and is based on it’s own scale. You can say this is analogous to a light sleeper saying even a clock disturbs him, but his wife can sleep thru police sirens. The RSSI scale and valuation is different for him and wife. So he might wake at 50/100, she might sleep thru everything but 60/80. Waking being an analogy to a signal loud enough to wake the transceiver that a signal has arrived that is understandable.
3. Antenna do nothing, but shape the power of the wifi signal. The regulations apply to the final shaped maximum transmission signal. For the USA, this is 26dBm. Most antenna shape the signal in a disk-like shape, so 20dBm (1watt sent thru antenna) is focused in a disc, this attains 6dB to the max 26dBm at the antenna, by concentrating the same power in a tighter volume.
4. In addition to the minimum amount of power is required to “wake” the transceiver, the most advanced radio modulation protocols require more power received to work. The more advanced encoding protocols, the more data that can be passed, per unit time. You can call this throughput. This is the Mbps value you see on the box. They are listing the theoretical maximum of the best modulation and signal strength it was engineered for.
5. The weaker the signal, the more likely noise will interfere with the data signal. The receiver will throw away data, it does not understand. This means bandwidth capacity is consumed by re-transmissions, and delays resulting from timeout before retransmission.
6. And finally, we get to distance. The obvious is the further you get, the weaker the signal from your wifi. Less obvious is you’re probably getting closer to sources of noise, making them louder, to your mobile wi-fi station. The signal loss, due to distance, in radio engineering is called Free Space Path Loss.

Free Space Path Loss

So obviously, you want to borrow wi-fi from a neighbor that lives nearby. But different materials, cause radio loss at different rates, and though loss thru distance thru the air is considerable, concrete and metal is worse. Open line of sight is the ideal situation, and is often the primary difference between indoor and outdoor published effective distance values. For the ideal line of sight situation, a back of napkin calculation for a reality test, can be calculated with what is called a link budget:

L-freespace can be obtained by a free space loss calculator. Which is how much theoretical loss between line of sight thru the air, in dB. This is signal strength loss, due to distance

P-recv is the power of signal, at the receiving wifi microchip(aka transceiver)
P-transmit is the power of signal, at the sending microchip(aka transceiver)
G-transmit is any amplification done, including power amplifier or sending antenna
L-transmit is any loss, from the sending transceiver, to the antenna, usually from signal line losses
L-freespace see above
G-recv is any amplification done, including power amplifier or receiving antenna
L-recv is any loss, from the receiving antenna, to the receiving transceiver, usually from signal line losses

Making sense of reported signal strength received, at client, from AP

P-recv from the formula, gets you a theoretical number that in no way resembles real life. With the formula, though, if you were to have real life numbers, there are things you can derive. For instance, 100ft for a laptop reported a -65dBm signal. But the Freespace Path loss (L-freespace) is -69.59dB, and for a AP transmitting at full power at 26dBm, the theoretical power received at the receiver, through this formula should be -27.59dBm.

P-recv = (P-transmit + G-transmit – L-transmit) – L-freespace – G-recv – L-recv
-65dBm = (___________max 26 dBm_______________) – 27.59dBm – (G-recv – L-recv – unknown)
-65dBm = (___________max 26 dBm_______________) – 27.59dBm – (40dB “fudge” consisting of G-recv – L-recv – unknown)

So I think it should be safe as a fudge factor, to subtract -40dB somewhere in between, to consider random factors you don’t know about.

Now you can make approximations. The signal strength allows more powerful modulations, which increase speeds

https://www.duckware.com/tech/wifi-in-the-us.html

https://morse.colorado.edu/~tlen5510/text/classwebch10.html

Interview-type answers

In the real world, these speeds translates to acceptable performance for the applications below.

In closing

So I use Wifi-n on 2.4Ghz, so Netflix can run on that, from 125ft away, with a directional antenna on the client wifi station. So performance can be acceptable.

It also needs to be noted that Wifi is a mutually exclusive medium on it’s spatial area that the signal covers. If your neighbor watches amazon prime, you are eating into his bandwidth way more, bc when he is closer, he spends less time transmitting bc the same data gets transmitted faster. Since you are further, you occupy the airwaves longer for same data, leaving less for him, if he decides to make facetime calls.

Also it should be noted, that if there is a 3rd neighbor who hates you for some reason, he might put his microwave (which is 2.45Ghz) in windows toward your antenna, and turn it on, all night, just to spite you.

These are things you need to know, if you plan to take borrowing too far.

PS
I’m using equipment you can find anywhere, to achieve effective distances up to the advertised limit of 300ft(90m). But these people in Canada, used specialty equipment w dishes to focus the signal, to achieve impressive distances.

https://www.youtube.com/watch?v=lYJFwXw1ZIc

I can’t verify that the equipment they used (or any that doesn’t just take a consumer wifi adapter and just change it’s antenna) is the regulated wifi frequencies of 2.4G, 5G, 6G, nor wifi modulations readable by most consumer devices. Meaning specialty equipment may change the frequency, or the modulations to achieve the distances, but pack your wifi data inside, and unpack during demodulation. So you may always need 2 of them. Basically, I don’t know if you can point your dish on the ground, and expect to connect to wifi with your iphone from 500ft away. This might be a good thing b/c who wants a condo highrise to pick up your SSID from km away.

Update: 8/10/2025, Starbucks Wifi obsession

So my obsession with borrowing Starbuck’s Wifi to tailgate, seems to show that actual throughput, is better reflected by the Link Quality value reported by the driver, which seems to be a arbitrary value that also takes into account the SNR, and/or measures the match (correlation) between the incoming DSSS signal and an ideal DSSS signal. And most unexpectedly, if a link is established, the average throughput doesn’t actually drop that much, with increased distance to AP. But there is wider variation in speeds, probably b/c of more errors. But AP-client negotiation doesn’t seem to change modulations, to lower the bitrate. Having no experience with Wifi drivers or real world performance of PHY layer, this went against my intuition as programmer(Layer 7, or 8 depending on your snarkiness). But Starbucks likes turning their wifi way up, and I can get a link from 150m away. I did not get approval from Starbucks to do this, but what they don’t notice, they will send the Wifi Police eventually for.

Starbucks wifi, is what every Post Office and Library should have had, as public infrastructure. That’s my compliment for the day. Please don’t send the Wifi police.

The signal level is usually a dBm value known as RSSI which is strength of signal it receives from AP, but my driver only reports a arbtirary value of 1 to 100. It doesn’t agree w the measured throughput, quite as well as the link quality metric

Below, is the roughly measured statistics, at distances away from a Starbuck’s AP. Thanks to Starbuck’s unpublished Store finder API, I was able to locate the approximate location of the Starbucks from where I was borrowing their signal, while having BBQ or drinks, watching Netflix over their Wifi

Distance vs. Measured Throughtput (m, est from GPS measurements vs. kbps)

Distance vs. Wifi driver reported Signal Level (m est from GPS measurements vs. x/100)

Distance vs. Wifi driver reported Signal Quality (m est from GPS measurements vs. x/100)

Thinking about it some more (and asking Google), it is likely that the 72.2 Mbps that my Wifi reports as the modulation, is likely Wireless-n on a single(out of 14, but in reality only 3) 2.4Ghz channel. That is way lower than the actual test 5Mbps that I get. This is very likely b/c whoever vendor supplies Starbucks w wifi, rate limits each client to 4Mbps. They each probably have LAP(Lightweight Access Point), and have a WLC at their WAN/ISP defining the capture page and client rate limits.

I have been able to connect to Starbuck’s Wifi, from 130m, and I suspect I can connect from 180m with free line of sight.