[srslte-users] Cell load estimation using srsLTE

[Falkenberg, Robert]

Hello Justin,

I went through the archives and found your message.
Passive LTE cell load estimation is exactly one of my research topics.
One approach is to decode resource allocations from all active users in the
cell and use machine learning to perform a cell load estimation or data rate
prediction [1].
Most resources are individually assigned to the UEs via PDCCH. (However,
resources may also be assigned via semi-persistent scheduling or via EPDCCH
if supported)
I have published a software "FALCON" -- which is built on top of the srsLTE
library -- that can decode resource assignments from PDCCH in real-time:
https://github.com/falkenber9/falcon

In short:
The main challenge of PDCCH decoding is the discovery of C-RNTIs of the
active UEs. Knowledge of RNTI is mandatory to validate the decoded data
structures [2].
If a UE joins a cell, its initial C-RNTI assignment can be extracted from
RAR [3].
To recover RNTIs of UEs that joined the cell before the observation period
one can perform statistical validation [2] or re-encoding [3, 4]. While the
first approach introduces a detection latency, the latter produces numerous
false-positives in non-ideal radio conditions (e.g. in presence of
interference, measurements at the edge of a sector/cell).

FALCON combines statistical validation, RAR tracking and a novel approach
called "shortcut decoding" to achieve fast discovery in non-ideal radio
conditions. (The corresponding research paper is currently under review).

Please don't hesitate to contact me, if you have further questions.

Best regards,
Robert

[1] R. Falkenberg, K. Heimann, C. Wietfeld, "Discover Your Competition in
LTE: Client-Based Passive Data Rate Prediction by Machine Learning", In
GLOBECOM 2017 - 2017 IEEE Global Communications Conference, Singapore, pp.
1-7, December 2017.
https://www.kn.e-technik.tu-dortmund.de/.cni-bibliography/publications/cni-p
ublications/Falkenberg2017c.pdf

[2] R. Falkenberg, C. Ide, C. Wietfeld, "Client-Based Control Channel
Analysis for Connectivity Estimation in LTE Networks", In IEEE Vehicular
Technology Conference (VTC-Fall), Montréal, Canada, September 2016.
https://www.kn.e-technik.tu-dortmund.de/.cni-bibliography/publications/cni-p
ublications/Falkenberg2016.pdf

[3] Nicola Bui, Joerg Widmer, "OWL: a Reliable Online Watcher for LTE
Control Channel Measurements", In: The 6th Workshop on All Things Cellular:
Operations, Applications and Challenges, The 22nd Annual International
Conference on Mobile Computing and Networking (ACM MobiCom 2016), 3-7
October 2016, New York, USA.
http://eprints.networks.imdea.org/id/document/4265

[4] LTE Radio Analytics Made Easy and Accessible, Swarun Kumar, Ezzeldin
Hamed, Dina Katabi, and Li Erran Li, ACM SIGCOMM 2014, Chicago IL.
https://www.andrew.cmu.edu/user/swarunk/papers/lteye-sigcomm2014.pdf

--
Robert Falkenberg, M.Sc.
Research Assistant

TU Dortmund University
Department Electrical Engineering & Information Technology
Communication Networks Institute
Otto-Hahn-Str. 6
44227 Dortmund
Germany

Phone: +49 231-755 6137
Fax: +49 231-755 6136
robert.falkenberg at tu-dortmund.de
www.kn.e-technik.tu-dortmund.de


--------------------------------
>From junione1997 at gmail.com  Fri Jun 14 04:04:19 2019
From: junione1997 at gmail.com (Justin Choi)
Date: Fri, 14 Jun 2019 14:04:19 +1000
Subject: [srslte-users] Cell load estimation using srsLTE
Message-ID:
<CAPcNpHHEpgTx8fNKbJzCEKfgUewLaPP+NzuK+fnYCQpRnMHU-w at mail.gmail.com>

Hello everyone,

I am currently working on a LTE cell load estimation project which should
passively identify the number of active users and throughput.
I am aware that there is a product called 'airscope' from srsLTE which does
this already but I want to try making my own version of it.
I have looked at multiple papers about this topic but I cant seem to find a
useful hint about how to achieve this.

Does anyone know how air-scope calculates the number of connected UEs and
data throughput?
If I am correct, these info are not broadcasted publicly by the cell and
therefore it can only be 'estimated'.
Any guidance would be greatly appreciated.

Kind regards
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 7856 bytes
Desc: not available
URL: <https://lists.srsran.com/pipermail/srsran-users/attachments/20190625/7c9cd743/attachment.bin>