Multilink: accuracy scheduling

[CheMax]

Hello,

I would like to learn a little more about the possibilities of thin planning of connection events. I rely on chapter 15, in particular on part 15.7 of the S140 SoftDevice specification v2.1.

In the particular case, I need to provide three simultaneous connections with different traffic levels. Sensors must provide real-time data transmission with a frequency of 10 Hz.
The radio is used at a speed of 1M, MTU is 247. Based on these data, I received the following requirements (the calculations in the table are estimates):

Sensor	Data Rate, byte per 100 ms	time to transfer, ms	Estimated Event Length
1	2000	33.75	27
2	200	3.75	3
3	25	2.5	2

Note: In the calculations, I rounded up the transfer time to the value of 1.25 ms, since the exchange is made in any case only within the double timeslot.

It is also worthwhile to understand that there is nothing ideal in this world, so situations are possible when the device cannot transmit data. Therefore, damping of data is provided before sending, followed by catch-up. Based on this, it will be more optimal to select the value of the GAP_CONNECTION_EVENT equal to 10. And allow the option to expand the connection interval. Then with more data available, I can get a picture roughly similar to this one from the specification:

More precisely for my case, this picture should look like this:

Where dark color indicates the transmission of data in a normal situation and the lightest possible time to extend the interval.

With a task more or less clarified, now to the questions:

Q1. Obviously, to ensure the planned value of the data flow, the connection sequence is necessary, as indicated in the figure. Otherwise, a device with a large amount of data will not have enough time to transfer data. Is redundancy manual positioning of connections possible?

Q2. If not, then at the application level, you can try to start in the correct sequence. BUT: if a device or two were turned off and then back on, how will the scheduler behave? Will he insert a new connection into an arbitrary place? That is, how to avoid, for example, such a situation:

Q3. Regarding the choice of the duration of the connection event: should this value be the same for both the master and the slave? The MTU setting affects, as the values are matched. and here?

Q4. Should the option for permission to extend the connection interval be active for both the master and the slave? Or is it enough just to allow the master?

Q5. In the options, it is possible to control an external amplifier, namely, activate it through a pin gpio. This can be used as an indicator of radio activity when debugging a multi-connection. Or are there any pitfalls?

---

Best regards,

amateur in asking difficult questions, Max

[Marjeris Romero]

Hi Max,

A1& A2. Could you explain what you mean with redundancy manual positioning? I didn't quite understood the question.

A3. For changing the connection event length you should set the right value for NRF_SDH_BLE_GAP_EVENT_LENGTH for both master and slave. You could also enable connection event length extention.  The connection event length doesn't need to be matched between master and slave devices, but it's recommended that all connection have intervals that have a common factor, see more information in https://infocenter.nordicsemi.com/topic/sds_s132/SDS/s1xx/multilink_scheduling/suggested_intervals_windows_s132.html?cp=4_6_2_0_14_9

A4. If the NRF_SDH_BLE_GAPP_EVENT_LENGTH is too low, then yes you should enable connection event length extension.

A5. yes, if you use PA/LNA, there is a own configuration you can use, see this blogpost for more information: https://devzone.nordicsemi.com/nordic/nordic-blog/b/blog/posts/pa-lna-support-in-s132

Best regards,

Marjeris

[CheMax]

Hi, 

I apologize for the long wait.

Marjeris Romero said:

A1& A2. Could you explain what you mean with redundancy manual positioning? I didn't quite understood the question.

I will try to explain in more detail.

I want to ensure a strict sequence of sensor connections in order to maximize the effective use of radio. Ideally, if I get the connections in the following order: yellow, green, blue (as in the picture above).This is not difficult to do at the application level if all three devices are already on and ready to go. Ok.

Now about the difficult. If for example, now only one or two devices are on the air. Can I somehow reserve a place in the temporary layout of the stack? To create a new connection at the time I need.

Best regards, Max

[Sigurd]

CheMax said:

If for example, now only one or two devices are on the air. Can I somehow reserve a place in the temporary layout of the stack? To create a new connection at the time I need.

A best-fit algorithm is used to find which free time space should be used when creating a new connection. This is explained and illustrated in the Scheduling Initiator timing chapter in the SDS. See this link.