[srslte-users] New numerology in srslte

[Zhouyou Gu]

Thanks Justin,

I made it working after I select smaller number of RBs.

Also, I found the time advance command at srsENB side is in microsecond.
srsenb/src/phy/prach_worker.cc
<https://github.com/srsLTE/srsLTE/blob/bf1982e28d0594a2a173a501ff9b1e22fbac442a/srsenb/src/phy/prach_worker.cc>
138
<https://github.com/srsLTE/srsLTE/blob/bf1982e28d0594a2a173a501ff9b1e22fbac442a/srsenb/src/phy/prach_worker.cc#L138>
if (prach_offsets[i] * 1e6 < max_prach_offset_us) {
139
<https://github.com/srsLTE/srsLTE/blob/bf1982e28d0594a2a173a501ff9b1e22fbac442a/srsenb/src/phy/prach_worker.cc#L139>
stack->rach_detected(b->tti, cc_idx, prach_indices[i],
(uint32_t)(prach_offsets[i]
* 1e6));
According to this page, the time advance command should be in 16 * Ts,
which is roughly around 0.5 us
http://howltestuffworks.blogspot.com/2014/07/timing-advance-and-time-alignment-timer.html


Is this a bug or my understanding is wrong.

Best Regards,
Zhouyou Gu


On Thu, 2 Apr 2020 at 19:33, Justin Tallon <
justin.tallon at softwareradiosystems.com> wrote:

> Hey Zhouyou,
>
> what rate are you using and what number of PRBs?
> ____
> Justin Tallon Ph.D.
>
> Software Radio Systems (SRS)
> http://www.softwareradiosystems.com
>
> +353-86-067-0753 | justin.tallon at softwareradiosystems.com
>
>
> On Wed, Apr 1, 2020 at 2:07 PM Zhouyou Gu <guzhouyou at gmail.com> wrote:
>
>> Hi Justin,
>> Thanks for your prompt reply!
>>
>> It was not in performance mode and I have changed it to performance for
>> srsUE and srsENB, as the following link
>> https://askubuntu.com/questions/604720/setting-to-high-performance
>>
>> There are still some UUULLL messages.
>> Is there any other tricks may help? :)
>>
>> Thanks again!
>>
>> Regards,
>> Charles
>>
>>
>>
>> On Wed, 1 Apr 2020 at 22:50, Justin Tallon <
>> justin.tallon at softwareradiosystems.com> wrote:
>>
>>> Hey Zhouyou,
>>>
>>> Are you sure your processor is running in PERFORMANCE mode?
>>>
>>> ____
>>> Justin Tallon Ph.D.
>>>
>>> Software Radio Systems (SRS)
>>> http://www.softwareradiosystems.com
>>>
>>> +353-86-067-0753 | justin.tallon at softwareradiosystems.com
>>>
>>>
>>> On Wed, Apr 1, 2020 at 12:23 PM Zhouyou Gu <guzhouyou at gmail.com> wrote:
>>>
>>>> Thanks a lot ! Justin,
>>>>
>>>> Actually, I just made it working with subcarrier spacing at 30kHz,
>>>> I made changes at the subcarrier spacing just like what you suggested.
>>>>
>>>> Two srsUEs can connect to the srsENB and I can ping them.
>>>>
>>>> Now the issue is that I found many UUULLL at srsUE
>>>> I think it is because the TTI is shorter and the CPU is not faster
>>>> enough to process the stack.
>>>>
>>>> I am wondering how can I improve this. What I am trying now is
>>>>
>>>>    - Try different CPU scheduling policy, e.g., SCHED_FIFO to
>>>>    SCHED_RR for DSP threads, Will this change help (I am not so familiar with
>>>>    CPU scheduling)?
>>>>    - Try allocate a fixed CPU mask for DSP.
>>>>    - Or the only solution is to get a better CPU that can be
>>>>    overclocked,
>>>>
>>>> My CPU is @3.4Ghz and Ubuntu has been configured with a low-latency
>>>> kernel.
>>>>
>>>> Thanks a lot!
>>>>
>>>> Regards,
>>>> Zhouyou Gu
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> On Wed, 1 Apr 2020 at 20:29, Justin Tallon <
>>>> justin.tallon at softwareradiosystems.com> wrote:
>>>>
>>>>> Hey Zhouyou,
>>>>>
>>>>> If you want to play with the subcarrier spacing my advice would be to
>>>>> pick a number of subcarriers you are happy with and then change the
>>>>> sampling rate at which it runs.
>>>>>
>>>>> The easiest way this could be done is with a scaling parameter,
>>>>> scaling the sampling rate by some factor. You will also need to change the
>>>>> timing being sent to your RF frontend as this is passed in milliseconds,
>>>>> not samples so this will also need to be scaled.
>>>>>
>>>>> In addition, pay attention to anywhere the subcarrier spacing (15000)
>>>>> is used explicitly, especially in CFO calculations.
>>>>>
>>>>> Regards,
>>>>> Justin
>>>>> ____
>>>>> Justin Tallon Ph.D.
>>>>>
>>>>> Software Radio Systems (SRS)
>>>>> http://www.softwareradiosystems.com
>>>>>
>>>>> +353-86-067-0753 | justin.tallon at softwareradiosystems.com
>>>>>
>>>>>
>>>>> On Tue, Mar 31, 2020 at 5:24 PM Zhouyou Gu <guzhouyou at gmail.com>
>>>>> wrote:
>>>>>
>>>>>> Dear srsLTE team,
>>>>>>
>>>>>> I am trying to implement a shorter TTI in srsLTE
>>>>>> My platform is B210 and dell computer.
>>>>>>
>>>>>> To implement the shorter TTI, I have changed the subcarrier spacing
>>>>>> to 30kHz (x2 than the original) and doubled the subcarrier spacing for
>>>>>> PRACH as well.
>>>>>> In addition, I have changed the clock rate setting.
>>>>>>
>>>>>> After I did this, srsUE cannot find the cell.
>>>>>>
>>>>>> What do you think about this idea?
>>>>>> Is it possible on my platform (B210).
>>>>>>
>>>>>> If yes, what else may also need to be changed?
>>>>>>
>>>>>> Thanks a lot!
>>>>>>
>>>>>> Regards,
>>>>>> Zhouyou
>>>>>>
>>>>>> _______________________________________________
>>>>>> srslte-users mailing list
>>>>>> srslte-users at lists.softwareradiosystems.com
>>>>>> http://www.softwareradiosystems.com/mailman/listinfo/srslte-users
>>>>>>
>>>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.srsran.com/pipermail/srsran-users/attachments/20200405/b181b5d0/attachment.htm>