TWT work in alignment with the application running on the device

Quoted from Heidi’s blog post Target Wake Time on the nRF7002 DK, published 23 Feb 2023 (available at https://devzone.nordicsemi.com/nordic/nordic-blog/b/blog/posts/target‑wake‑time‑on‑the‑nrf7002‑dk): “TWT is expected to work in alignment with the application running on the device. This means the application should predict the traffic, and make sure that it can be blind to the network for the period for which it is sleeping.”

How does Nordic handle TWT working in alignment with the application running on the device?

Parents
  • Hi ChengQing Zhao,

    According to what is mentioned in the blog post, you have to use intervals which suit your application. If you have a requirement for response time from for example Cloud, then TWT can add a significant delay here. However, as long as you have an application which sends data (and thus never receives from Cloud/network with requirements for response time), you can use TWT. 

    Please let us know if you have any further questions. 

    Best Regards,

    Samruddhi

  • Hi Samruddhi,

    Thank you for your detailed response.

    I understand that selecting appropriate TWT intervals based on application requirements works when the application's traffic pattern is known in advance. However, in practice, traffic patterns are often unknown or change dynamically.

    My main concern is Heidi's statement that "the application should predict the traffic." I also noticed that Nordic is actively promoting Edge AI on its official website.

    This leads to my question: Has Nordic explored using Edge AI to predict application traffic patterns and dynamically adapt TWT parameters, so that TWT can work in alignment with the application?

    If so, are there any examples or implementations available for the nRF7002? If not, how does Nordic recommend handling unpredictable traffic while maintaining the energy-saving benefits of TWT?

    Thank you again for your support.

    Best regards

Reply
  • Hi Samruddhi,

    Thank you for your detailed response.

    I understand that selecting appropriate TWT intervals based on application requirements works when the application's traffic pattern is known in advance. However, in practice, traffic patterns are often unknown or change dynamically.

    My main concern is Heidi's statement that "the application should predict the traffic." I also noticed that Nordic is actively promoting Edge AI on its official website.

    This leads to my question: Has Nordic explored using Edge AI to predict application traffic patterns and dynamically adapt TWT parameters, so that TWT can work in alignment with the application?

    If so, are there any examples or implementations available for the nRF7002? If not, how does Nordic recommend handling unpredictable traffic while maintaining the energy-saving benefits of TWT?

    Thank you again for your support.

    Best regards

Children
No Data
Related