I am trying to build a video streaming pipeline using WiFi HaLow as the transmission medium. However, the distance I am able to achieve seems much less than what the videos the chip companies showcase.
(for example this youtube video posted by morsemicro shows that 3km video streaming range is possible: https://www.youtube.com/watch?v=fBMgZah2Z7g)
Meanwhile for the test I did, I am only able to achieve around 300m stable video stream along the East Coast Park beach in Singapore.
I am using a pair of Heltec HT-H7608 wifi halow devices (which should use the MM6108 chip). At around 300m, my video stream cut off and the connection between devices on the two ends became very unstable. I could still ping the other device at this distance but typing letters in the terminal after ssh into the other device became extremely difficult. The signal/noise was about -75dBm/-80dBm.
I am curious about what are the factors that might affect the results? (Like transmit power etc. ?) What is the ideal signal/noise stats that I should achieve? And what are the possible ways for me to boost the range further?
Couple questions. What is the channel width you are using: 1/2/4/8? Also do you know the bitrate of the video you are trying to pull? Maybe you are just trying to push too much video? Have you done any iperf testing using UDP to see what your throughput is? Maybe there are other optimizations that can be done there prior to trying to pull a video feed.
The closer the signal strength gets to the environment noise the less you’ll be able to transmit.
Good call from @castironclay, can you check what channel the AP is operating on? And the raw throughput you can achieve at 300m range.
Do you know if the STA has been configured to have power save enabled? This may increase the latency experienced if the STA goes to sleep between key strokes
I am using 4MHz channel width (the max I can have for SG region);
I didn’t manage to do an iperf3 test at the beach. But when I did a test near the office, I could get about 600Kb/sec throughput when the signal/noise ratio is about -70dBm/-80dBm.
What I am curious about is how does the stats of a healthy connection look like (especially when it’s at distance). For example, do I get -50/-80dBm or -70/-100dBm at long distance? And what is a healthy throughput at long distance from wifi halow? If I can have a reference, it would be easier for me to debug.
Revisiting this thread and noticed you had some unanswered questions @alexgosleep , hopefully it’s not too late to give you some answers
What I am curious about is how does the stats of a healthy connection look like (especially when it’s at distance). For example, do I get -50/-80dBm or -70/-100dBm at long distance? And what is a healthy throughput at long distance from wifi halow? If I can have a reference, it would be easier for me to debug.
Disclaimer:
This is a very hard thing to quantify as it depends heavily on the environment. For example, depending on the profile of the noise you may have a low noise floor but one which triggers collision sense such that the Wi-Fi radio has to back off for a short period of time, which will impact throughput.
Your best bet will always be to use a spectrum analyzer to capture what the environment looks like, and select the appropriate channel/bandwidth for your environment. Practically the entire business model for companies like Ekahau
The easiest component to quantify here is the signal strength (first number). This is a function of distance and can be modeled relatively easily with the Friis equation. There are plenty of calculators online which will tell you the power at the receiver for a given frequency, distance, and transmit power. See for example https://www.pasternack.com/t-calculator-friis.aspx.
This of course assumes there is not obstruction to the signal. Any material blocking line of sight will reduce signal strength further - though S1GHz is generally reasonable for material penetration compared to higher frequency technologies.
You also want to consider the fresnel zone. Ensure your antennas are high so as to avoid having too much of the fresnel zone impeded.
The next component to quantify is noise (second number). This is something you can’t control directly as it is largely environmental but some of the things which impact noise are:
Other intentional in-band transmitters, as well as adjacent band transmitters raising the RF noise floor
A wider channel width causes the receiver to listen to a wider “slice” of the band, and therefore “collect” more noise power
Unintentional in-band transmitters (eg, devices poorly designed for EMC) radiating power in band. Think industrial motors and such.
Your noise at the transmitter may differ from the noise at the receiver! And it is impossible for us to tell you what the measured noise will be.
Focusing on the theory only. The higher your signal to noise ratio, the better rate your devices can talk at. I gave a run-down of the mathematics behind this here:
