Hello there
I do not have iphone, do I want to use Raspberry indoor positioning
I want to use RSSI to measure the distance, but found that the measured conversion per meter distance error is large, may I ask how did you measure?
formula:
d = 10 ^ ((P-Rssi) / 10n) (n ranges from 2 to 4)
This is most commonly seen rssi change the distance formula
The formula you’re using is indeed one of the most popular (slightly more about where it comes from in this post: Determine accurate distance of signal)
From now on you have everything you need to calculated distance. P in your formula is TxPower which beacons are transmitting as part of the package and Rssi you should get from your device. The most common n I’ve seen is 2-2.5.
As of accurate distance that’s longer story. Beacon signal is quite “unstable” and you’ll get a lot of jumping of values. So you should probably take a few values and average them over time, i.e. over last second (so called “smoothing”). And still if you have a lot of interference in your area, it’s still gonna be a bit off. Beacons are not great for calculating accurate distance. They’re more about saying you’re in close proximity or not of some object.
Hi
Thank you, I understand
But the beacon to the TxPower
I actually different Rssi values measured one meter
It is recommended to use the actual measurement or beacon provided TxPower
Beacon TxPower: -60dbm
The actual measured 1 meter: -47dbm
Yeah, you’re not gonna get perfect results. It’s not very precise. Using higher broadcast power should give you better accuracy at short distances, but it won’t make thing drastically better.
Because I want to know the track and the distance between the target
but… now
Why I can not adjust the release time interval
advertising Interval: 953ms
I’m just curious what do you want to use the distance for? Is there any specific use case you have in mind (e.g., you want to open the door with your Raspberry Pi when somebody gets close enough to it), or are you simply playing the technology, exploring what’s possible?
Well, yes, but why do you need the location? Is the Raspberry Pi attached to something, a robot you want to steer and thus you need to know its position? We’ll be happy to help you all the way we can, but we need to know what you’re trying to do
Knowing the location is just a tool, but you surely have something specific you want to use this tool, the location, for, and that’s where we might be able to help—suggest a better alternative, etc.
As for the more immediate problem, the interval—which app is that? Estimote iOS? Estimote Android?
Thanks Piotr
I want to use is positioned within a space in which people can know the position, and then the means to achieve control, such as lights, doors, fans … and so on, so from Beacon achieve a positioning function.
Home automation, that’s a nice idea! But uh, you want to be carrying Raspberry Pi with you at all times? (:
That’s rather impractical, I think using your phone for indoor positioning would be better. You said you don’t have an iPhone, do you have an Android phone? You can try our Proximity app template to see how you can detect when somebody moves from one beacon to another:
You could use that to control the home appliances, e.g., if the app detects you moved from the beacon in the kitchen to the beacon in the living room, turn the lights off in the kitchen, etc.
To give this a legitimate answer there are a few steps you need to take to create an accurate localization system using these beacons.
The first thing you’ll need is to develop a path loss model, this is similar to what you’re doing in the original post. A pathloss model can help calculate distances based on transmit power and Rssi (received signal strength).
Once you have a solid pathloss model (you’ll need to do empirical testing with these devices to help find the shadow fading), you’ll need to adopt an algorithm. This could be trilateration, least means squares, maximum likelihood, there’s many algorithms to choose from. Essentially you’ll be calculating distances from your three points and using these distances in your algorithm to determine a point of localization.
For simplicity to start I’d recommend trilateration or Least Means with channel modeling:
Here’s where it gets interesting. You want extreme accuracy? That’s actually cutting edge research at the moment, everyone is hiring for indoor localization google, amazon, estimote, etc. So if you can figure this out you’ve got a pretty guaranteed job.
To give you a starting point and some search terms, you’re going to need a probability/statistical analysis associated with your algorithm. One example is Cramer Rao lower bound. When using estimated distances, which is what your pathloss model is calculating, we can use statistical analysis to determine a maximum likelihood of certain point to be accurate. This would require you to take empirical measurements within the building you’re trying to localize. Basically you find points where your accuracy is very good to all 3 beacons or access points. You can calculate centroids based off of this data set and create a model to determine probabilities of accuracy.
I understand this is all gibberish without learning the basics behind it. I’m in the process of finishing a paper on this exact topic that has lots of great examples and will link it here when I’m done.