Problem getting AoA from CSI data

[minseuk]

Hello. I'm a student studying Wi-Fi indoor localization using CSI packets. I got CSI packets with following csitool and got Unwrapped CSI phases. After ToF sanitization, I can obtain AoA values, but the result is different from actual angle in the environment. I did experiment on 5.745GHz and 3cm antenna distance.

Acutally, I studied this csitool for a long time, and I think my analysis algorithm including sanitization and MUSIC is correct. I guess the problem is caused from CSI data.

For example, this is a data for 60 degree, but it shows the AoA is 30 degree (have to add -90 degree in the pseudospectrum graph)

Is there any issue such as 'phase reverse' or 'x+90 degree x-90 degree' from your experience? I don't know if it was coincidence, but when i added Pi for certain antenna phase, I got almost exact result. I expect your response.

[sandeep-bastola]

Hello Minseuk,

I am also a student working on wifi based indoor localisation. But since I dont deal with AoA in my research, I cannot provide much inputs here. But sure we can help each other in many respects in the topic. Feel free to communicate me at ga42gup@tum.de . Thanks,

[SusieYonng]

Maybe your ground truth is 30 degrees? and keep the antenna spacing no more than wavelength/2.

[minseuk]

Hello SusieYonng. For this example, the phase difference from each antenna is about 0.35*2pi. I did an experiment in 5.745GHz and the 3cm antenna distance, so mathematically It shows asin(0.35/0.5745) = 37 degree. But the ground truth was 60 degree.

[lax7]

Hi Minseuk, can you share your email. We can discuss, I also have similar problem.

[esoltana]

There is a constant LO offset between the three antennas that need to be manually removed before estimating AoA. Please check ArrayTrack paper for more information and the calibration process.

[minseuk]

@laxima Hello laxima. I'm sorry for late answer. My email is kms4105@kaist.ac.kr Thank you.

[minseuk]

@soltanaghaei Hello soltanaghaei. Thank you for your advice. I found ArrayTrack paper and know how to solve the offset problem. It was represented in 'implementation' part of that paper.

[minseuk]

Now, my next problem is how to handle 'differ phase difference' for each subcarrier. As you can see, the phases of each antenna and subcarrier is varying. There must be multi-path problem and i'm searching the way if i can subtract several subcarriers that are considered as single path phase, or properly add or subtract phases. I'm currently reading the paper Tzur, Asaf, Ofer Amrani, and Avishai Wool. "Direction Finding of rogue Wi-Fi access points using an off-the-shelf MIMO–OFDM receiver." Physical Communication 17 (2015): 149-164.

[titinho11]

Hello Minseuk,

I am also a student working on wifi based indoor localisation system, and i am also facing inconsistant angle of arrival results. Does Arraytrack's calibration make it work for you ?

[zwlovefish]

@minseuk hello, i am also dealing with localization system,Could you send me your matlab code ？please help me. thank you

[zwlovefish]

@minseuk my email is zjw1149610059@gmail.com