NCS AOA: IQ sample data type.

[Harvey Zheng]

Hi,

I am using Zephyr3.0.0 to validate AOA CTE sampling. 

Firstly,  as below is the official data format. Nordic Semiconductor Radio peripheral provides 16 bit wide IQ samples and the first 4 bits are the sign bit extension.

Secondly, BT 5.3 Core specification Vol 4, Part E sectons 7.7.65.21 and 7.7.65.22 limit size of IQ samples to 8 bits. So in this commit, Zephyr right shift the data 4bits.

I printed out some IQ sample data as below.

This is my print information: 2112.df.zip

I have some questions about this. It seems to missing some valid bits of data?  How will the loss of these data bits affect the positioning accuracy? Or does the application need to do something special with these IQ datas such as left shift them 4 bits?

[Simonr]

Hi

The AoA direction finding solution we provide will only output the raw IQ data by default, as we don't provide the AoA algorithm necessary to make use of the data collected by the antenna array. There is no point in looking at the raw data as it needs to be converted into vector form to review it properly. You can check out this page for information on the various formulas used to translate IQ data between polar and rectangular form.

Best regards,

Simon

[Harvey Zheng]

Hi，

I noticed that by shifting right 4 bits, the data 0xFFC7 becomes 0xFFFC. It can be interpreted as low 4 bits are useless data?

Thanks.

[Simonr]

Can you provide some more details on what exactly you're doing? What bits are you right shifting exactly? Are you still trying to look at the raw IQ data?

Best regards,

Simon

[Harvey Zheng]

Hi，

I try to look at the raw IQ data, because I want to see the difference between the IQ data in cte_report_cb and the value in the Nordic register.

I printed the IQ data here: https://github.com/zephyrproject-rtos/zephyr/blob/zephyr-v3.0.0/subsys/bluetooth/controller/hci/hci.c#L2845

0xFFC7 is the data from register，but it becomes 0xFC in CTE receive callback. That's why I think it miss some valid bits after right shift.

[Simonr]

Hi

As I mentioned earlier, you won't get reliable data from raw IQ data, so I don't recommend using time/resources on this. You need to convert the IQ data to vector form to get useful data, and then work on any issues from there.

Best regards,

Simon

[swz_nordic]

ok,thanks.I will try your suggession.but now i see that i can use the iq data to get phase difference,then using it to calclulate  the anglr of the receiver.i think it is easier.