Private LTE with LimeSDR and srsRAN – Part 4 (Config & Launch)

|

Introduction

If you managed to land here without seeing part one, part two and part three in my series on setting up my own private LTE network, you may want to check those out first.

Continuing from where we left off, having programmed our SIM card. The next step involves configuring srsRAN and firing up the network. Before connecting our first mobile device to it.

This post therefore should if all goes well complete the initial series and leave us with our goal of a private and useful LTE network.

Configuration

In order to configure srsRAN we need to tweak the files found in ~/.config/srsran (which were added by the srsran_install_configs.sh script when building).

For example my directory currently looks like this:

user@deblaptop:~$ ls -l $HOME/.config/srsran/
total 72
-rw-r--r-- 1 user user 18362 Dec 22 15:28 enb.conf
-rw-r--r-- 1 user user  3796 Dec 22 15:28 epc.conf
-rw-r--r-- 1 user user  1688 Dec 22 15:28 mbms.conf
-rw-r--r-- 1 user user  3386 Dec 22 15:28 rb.conf
-rw-r--r-- 1 user user  2951 Dec 22 15:28 rr.conf
-rw-r--r-- 1 user user 10584 Dec 22 15:28 sib.conf
-rw-r--r-- 1 user user 19111 Dec 22 15:28 ue.conf
-rw-r--r-- 1 user user  1275 Dec 22 15:28 user_db.csv

enb.conf

The enb.conf file configures the eNodeB component. In LTE terms the eNodeB is the hardware / software component responsible for communicating with the mobile devices. These live inside cellular base stations. It will manage the SDR radio, communicating with the mobile devices. While also communicating with the EPC (evolved packet core), described below.

In the “[enb]” section we need to update the MCC and MNC, with those selected during SIM card programming. My MCC is 999 and MNC is 69, therefore the section now contains:

[enb]
enb_id = 0x19B
mcc = 999
mnc = 69
...

In the “[rf]” section we need to inform srsRAN that we’re using a LimeSDR, while selecting which antenna ports are in use. We use the device_name and device_args parameters for this.

We also need to set the transmit and receive gain which should be applied to the LimeSDR via the tx_gain and rx_gain paramters. I’ve set the gains here based on a LimeSDR forum post I happened across on my travels. It seems to work ok(ish) for me, although may need some tweaking.

My updated “[rf]” section is therefore as follows:

[rf]
#dl_earfcn = 3350
tx_gain = 66
rx_gain = 47

device_name = soapy
device_args = driver=lime,rxant=LNAH,txant=BAND2
...

rr.conf

Finally we need to select the radio frequency to transmit and receive on, which can conveniently be expressed via a EARFCN (Evolved-UTRA Absolute Radio Frequency Number). srsRAN only supports FDD (frequency division duplexing), whereby the downlink and uplink from the mobile device occurs on different frequencies. By providing the downlink EARFCN number, srsRAN is able to calculate the downlink frequency. Before using its knowledge of the LTE bands and uplink / downlink offsets to calculate the uplink frequency. I used this neat calculator to select an EARFCN number at the start of LTE Band 7 which supports the full 20Mhz bandwidth, yielding a download EARFCN of 2850.

We could update the downlink EARFCN in the enb.conf file using the dl_earfcn which looks to be an override value. I believe the correct location to set the downlink channel is in the rr.conf file.

Find the cell_list section and update the dl_earfcn parameter. The section of my updated file is as follows:

...
cell_list =
(
  {
    // rf_port = 0;
    cell_id = 0x01;
    tac = 0x0007;
    pci = 1;
    // root_seq_idx = 204;
    dl_earfcn = 2850;
    //ul_earfcn = 21400;
    ho_active = false;
...

With that the eNodeB configuration is complete.

epc.conf

The epc.conf file configures the EPC. In a commercial network the EPC consists of several nodes, providing different services. Services for example to authenticate users, to track usage, to track which eNodeB a user is connected to as well as transferring data in and out of the network. srsRAN has neatly combined these services into a single binary, which for a small test setup makes things super simple.

In the “[mme]” section we need to update the MCC and MNC as above. My “[mme]” section therefore appears as follows:

[mme]
mme_code = 0x1a
mme_group = 0x0001
tac = 0x0007
mcc = 999
mnc = 69

With that the EPC configuration is complete.

user_data.csv

The user_data.csv file contains the different SIM card identities that we configured earlier. It’s used by the HSS (home subscriber service), which is part of the EPC binary described above.

The file contains a few example identities which we can update / remove as appropriate.

The details we programmed into the SIM card are now required, specifically the IMSI, Ki and OPc values. With the notes at the top of the file providing a useful description of what each field means.

My updated file, based on my SIM configuration is as follows:

#
# .csv to store UE's information in HSS
# Kept in the following format: "Name,Auth,IMSI,Key,OP_Type,OP/OPc,AMF,SQN,QCI,IP_alloc"
#
# Name:     Human readable name to help distinguish UE's. Ignored by the HSS
# Auth:     Authentication algorithm used by the UE. Valid algorithms are XOR
#           (xor) and MILENAGE (mil)
# IMSI:     UE's IMSI value
# Key:      UE's key, where other keys are derived from. Stored in hexadecimal
# OP_Type:  Operator's code type, either OP or OPc
# OP/OPc:   Operator Code/Cyphered Operator Code, stored in hexadecimal
# AMF:      Authentication management field, stored in hexadecimal
# SQN:      UE's Sequence number for freshness of the authentication
# QCI:      QoS Class Identifier for the UE's default bearer.
# IP_alloc: IP allocation stratagy for the SPGW.
#           With 'dynamic' the SPGW will automatically allocate IPs
#           With a valid IPv4 (e.g. '172.16.0.2') the UE will have a statically assigned IP.
#
# Note: Lines starting by '#' are ignored and will be overwritten
ue1,mil,999690000000001,895fd70bd46fa42435ba1471665b0028,opc,9b9fb8dcae42ae23011abc20bb80796a,8000,000000000000,7,dynamic

The AMF number should be set to 8000 of above. From one of the 3GPP documents either the lowest order bit (or possibly highest order bit based on the examples above) is used to indicate that this device is connecting over LTE, rather than an alternative access technology such as Wi-Fi. The remaining bits are vendor specific and unused by srsRAN as far as I could tell.

The SQN is a sequence number which ensures the period before an authentication request for a device is repeated is sufficiently large to prevent replay attacks. We can set this to zero and it will be auto-incremented by the EPC.

QCI relates to the quality of service class for the user….given that we’d be happy with any service at present, leaving it at the example default of 7 seems reasonable.

With that all of the configuration is complete.

Startup

In order to start the network up ready for clients we need to launch the EPC followed by the eNodeB.

With the SDR connected and RF cables to the mobile device already in place.

Using the same prefix as was used when building and installing:

export SRSRAN_INSTALL=${HOME}/srsRAN

In one terminal launch the EPC:

sudo LD_LIBRARY_PATH=${SRSRAN_INSTALL}/lib sh -c "cd ${HOME}/.config/srsran; ${SRSRAN_INSTALL}/bin/srsepc epc.conf"

If all goes well it should output something similar to the following at startup:

Built in Release mode using commit 61059ecbc on branch release_23_04.


---  Software Radio Systems EPC  ---

Reading configuration file epc.conf...
HSS Initialized.
MME S11 Initialized
MME GTP-C Initialized
MME Initialized. MCC: 0xf999, MNC: 0xff69
SPGW GTP-U Initialized.
SPGW S11 Initialized.
SP-GW Initialized.

In another terminal launch the eNodeB:

echo "performance" | sudo tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor
sudo LD_LIBRARY_PATH=${SRSRAN_INSTALL}/lib sh -c "cd ${HOME}/.config/srsran; ${SRSRAN_INSTALL}/bin/srsenb enb.conf"

If all goes well it should output something similar to the following at startup:

Active RF plugins: libsrsran_rf_soapy.so
Inactive RF plugins: 
---  Software Radio Systems LTE eNodeB  ---

Reading configuration file enb_lime_soapy.conf...

Built in Release mode using commit 61059ecbc on branch release_23_04.

Opening 1 channels in RF device=soapy with args=driver=lime,rxant=LNAH,txant=BAND2
Supported RF device list: soapy limesdr file
Soapy has found device #0: addr=1d50:6108, driver=lime, label=LimeSDR-USB [USB 3.0] 908340187372A, media=USB 3.0, module=FX3, name=LimeSDR-USB, serial=000908340187372A, 
Selecting Soapy device: 0
[INFO] Make connection: 'LimeSDR-USB [USB 3.0] 908340187372A'
[INFO] Reference clock 30.72 MHz
[INFO] Device name: LimeSDR-USB
[INFO] Reference: 30.72 MHz
[INFO] LMS7002M register cache: Disabled
Setting up Rx stream with 1 channel(s)
Setting up Tx stream with 1 channel(s)
[INFO] RX LPF configured
[INFO] Filter calibrated. Filter order-4th, filter bandwidth set to 5 MHz.Real pole 1st order filter set to 2.5 MHz. Preemphasis filter not active
[INFO] TX LPF configured
Available device sensors: 
 - clock_locked
 - lms7_temp
Available sensors for Rx channel 0: 
 - lo_locked
Setting Rx channel 0 antenna to LNAH
Setting Tx channel 0 antenna to BAND2
State of gain elements for Rx channel 0 (AGC not supported):
 - TIA: 9.00 dB
 - LNA: 30.00 dB
 - PGA: -7.00 dB
State of gain elements for Tx channel 0 (AGC not supported):
 - PAD: 0.00 dB
 - IAMP: 0.00 dB
Rx antenna set to LNAH
Tx antenna set to BAND2

==== eNodeB started ===
Type <t> to view trace
[INFO] RX LPF configured
[INFO] Filter calibrated. Filter order-4th, filter bandwidth set to 11.52 MHz.Real pole 1st order filter set to 2.5 MHz. Preemphasis filter not active
[INFO] TX LPF configured
Setting frequency: DL=2630.0 Mhz, UL=2510.0 MHz for cc_idx=0 nof_prb=50
[INFO] Tx calibration finished
[INFO] Rx calibration finished

Finally to allow clients onto the internet we need to enable IP forwarding and configure a NAT rule for the system. Clients will be assigned an IP in the range 172.16.0.0/255 with their traffic emerging from a Linux TUN interface. The IP forwarding and NAT configuration will allow them to access the internet via the host’s default route.

This has been neatly wrapped up by the srsRAN team in a small helper script, which we can execute as follows. Substituting enp3s0f2 for the name of your internet connected Ethernet interface.

sudo ${SRSRAN_INSTALL}/bin/srsepc_if_masq.sh enp3s0f2

If all goes well it should output the single line (with the appropriate interface name of course):

Masquerading Interface enp3s0f2

The network should be online and serving clients.

First Client

If you’ve got this far, I’d hazard a guess that you’ll be capable of configuring your mobile device. I thought this about myself….until what shall be referred to as “modem firmware and old configuration gate” occurred.

All you should need to connect is the network’s APN name which is “srsran”.

My test setup is using a Sierra Wireless MC7455, connected to an Ubuntu machine which is using ModemManager. I can therefore issue commands such as the following to scan for networks:

phil@ubuntu:~$ mmcli -m any --3gpp-scan --timeout=60
  ---------------------
  3GPP scan | networks: 99969 - Software Radio Systems  (lte, current)

Woohoo, there it is!

Connecting should then just be a matter of executing something similar to the following:

phil@ubuntu:~$ mmcli -m any --simple-connect='apn=srsapn,ip-type=ipv4'
successfully connected the modem

phil@ubuntu:~$ journalctl -u ModemManager -n 10
-- Logs begin at Mon 2022-06-06 10:55:32 BST, end at Sun 2022-12-25 20:15:52 GMT. --
Dec 24 20:04:43 ubuntu ModemManager[181550]: [/dev/cdc-wdm0] Allocating new client ID...
Dec 24 20:04:43 ubuntu ModemManager[181550]: [/dev/cdc-wdm0] Registered 'wds' (version 1.67) client with ID '36'
Dec 24 20:04:44 ubuntu ModemManager[181550]: <info>  [modem14/bearer44] QMI IPv4 Settings:
Dec 24 20:04:44 ubuntu ModemManager[181550]: <info>  [modem14/bearer44]     address: 172.16.0.5/30
Dec 24 20:04:44 ubuntu ModemManager[181550]: <info>  [modem14/bearer44]     gateway: 172.16.0.6
Dec 24 20:04:44 ubuntu ModemManager[181550]: <info>  [modem14/bearer44]     DNS #1: 8.8.8.8
Dec 24 20:04:44 ubuntu ModemManager[181550]: <info>  [modem14/bearer44]     DNS #2: failed (Field 'Secondary IPv4 DNS Address' was not found in the message)
Dec 24 20:04:44 ubuntu ModemManager[181550]: <info>  [modem14/bearer44]        MTU: 1500
Dec 24 20:04:44 ubuntu ModemManager[181550]: <info>  [modem14] state changed (connecting -> connected)
Dec 24 20:04:44 ubuntu ModemManager[181550]: <info>  [modem14] simple connect state (8/8): all done

phil@ubuntu:~$ mmcli -m any --simple-disconnect
successfully disconnected all bearers in the modem

While that got the modem connected, as could be seen in the journal log it didn’t assign the IP we were provided to the interface. For that network manager is required. We can create a connection as follows:

sudo nmcli c add con-name "TestNET" type gsm ifname "*" apn "srsapn" connection.autoconnect no

Which can then be connected as follows:

sudo nmcli c up TestNET

Which if all goes well, leads to ModemManager connecting the modem and NetworkManager configuring the interface with the details returned:

phil@ubuntu:~$ ip address show dev wwan0
41: wwan0: <POINTOPOINT,MULTICAST,NOARP,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UNKNOWN group default qlen 1000
    link/none 
    inet 172.16.0.5/30 brd 172.16.0.7 scope global noprefixroute wwan0
       valid_lft forever preferred_lft forever

We should be able to ping the internet:

phil@ubuntu:~$ ping -I wwan0 8.8.8.8
PING 8.8.8.8 (8.8.8.8) from 172.16.0.5 wwan0: 56(84) bytes of data.
64 bytes from 8.8.8.8: icmp_seq=1 ttl=115 time=34.4 ms
64 bytes from 8.8.8.8: icmp_seq=2 ttl=115 time=31.7 ms
64 bytes from 8.8.8.8: icmp_seq=3 ttl=115 time=29.8 ms
64 bytes from 8.8.8.8: icmp_seq=4 ttl=115 time=28.6 ms
64 bytes from 8.8.8.8: icmp_seq=5 ttl=115 time=46.7 ms

And of course not forgetting the obligatory speed test, running iperf3 as a server on the SDR machine:

iperf3 -s

And on the machine with the modem. I add a specific route as the machine has an Ethernet card with a default route of its own which must be avoided. iPerf is run with the “-R” option for reverse, meaning the server sends and the client receives, aka the client is downloading from the server:

sudo ip route add 172.16.0.1 dev wwan0
iperf3 -c 172.16.0.1 -R -t 30

Here’s the result:

phil@ubuntu:~$ iperf3 -c 172.16.0.1 -R -t 30
Connecting to host 172.16.0.1, port 5201
Reverse mode, remote host 172.16.0.1 is sending
[  5] local 172.16.0.5 port 55018 connected to 172.16.0.1 port 5201
[ ID] Interval           Transfer     Bitrate
[  5]   0.00-1.00   sec  3.88 MBytes  32.5 Mbits/sec                  
[  5]   1.00-2.00   sec  3.36 MBytes  28.2 Mbits/sec                  
[  5]   2.00-3.00   sec  4.08 MBytes  34.3 Mbits/sec                  
[  5]   3.00-4.00   sec  3.20 MBytes  26.8 Mbits/sec                  
[  5]   4.00-5.00   sec  3.27 MBytes  27.4 Mbits/sec                  
[  5]   5.00-6.00   sec  4.06 MBytes  34.0 Mbits/sec                  
[  5]   6.00-7.00   sec  4.02 MBytes  33.7 Mbits/sec                  
[  5]   7.00-8.00   sec  4.09 MBytes  34.3 Mbits/sec                  
[  5]   8.00-9.00   sec  4.06 MBytes  34.1 Mbits/sec                  
[  5]   9.00-10.00  sec  3.61 MBytes  30.3 Mbits/sec                  
[  5]  10.00-11.00  sec  4.04 MBytes  33.9 Mbits/sec                  
[  5]  11.00-12.00  sec  4.03 MBytes  33.8 Mbits/sec                  
[  5]  12.00-13.00  sec  3.90 MBytes  32.7 Mbits/sec                  
[  5]  13.00-14.00  sec  3.96 MBytes  33.2 Mbits/sec                  
[  5]  14.00-15.00  sec  3.17 MBytes  26.6 Mbits/sec                  
[  5]  15.00-16.00  sec  3.95 MBytes  33.2 Mbits/sec                  
[  5]  16.00-17.00  sec  3.24 MBytes  27.1 Mbits/sec                  
[  5]  17.00-18.00  sec  4.08 MBytes  34.2 Mbits/sec                  
[  5]  18.00-19.00  sec  4.10 MBytes  34.4 Mbits/sec                  
[  5]  19.00-20.00  sec  4.00 MBytes  33.5 Mbits/sec                  
[  5]  20.00-21.00  sec  3.58 MBytes  30.0 Mbits/sec                  
[  5]  21.00-22.00  sec  3.16 MBytes  26.5 Mbits/sec                  
[  5]  22.00-23.00  sec  3.85 MBytes  32.3 Mbits/sec                  
[  5]  23.00-24.00  sec  3.99 MBytes  33.4 Mbits/sec                  
[  5]  24.00-25.00  sec  3.27 MBytes  27.5 Mbits/sec                  
[  5]  25.00-26.00  sec  3.97 MBytes  33.3 Mbits/sec                  
[  5]  26.00-27.00  sec  4.04 MBytes  33.9 Mbits/sec                  
[  5]  27.00-28.00  sec  4.11 MBytes  34.4 Mbits/sec                  
[  5]  28.00-29.00  sec  3.82 MBytes  32.0 Mbits/sec                  
[  5]  29.00-30.00  sec  4.03 MBytes  33.8 Mbits/sec                  
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Retr
[  5]   0.00-30.05  sec   116 MBytes  32.3 Mbits/sec   38             sender
[  5]   0.00-30.00  sec   114 MBytes  31.8 Mbits/sec                  receiver

iperf Done.

Around 32 Mbit/s, with a few retires. Overall not too shabby.

How about upload:

iperf3 -c 172.16.0.1 -t 30

Here’s the result:

phil@ubuntu:~$ iperf3 -c 172.16.0.1 -t 30
Connecting to host 172.16.0.1, port 5201
[  5] local 172.16.0.5 port 55842 connected to 172.16.0.1 port 5201
[ ID] Interval           Transfer     Bitrate         Retr  Cwnd
[  5]   0.00-1.00   sec  2.65 MBytes  22.2 Mbits/sec    0    140 KBytes       
[  5]   1.00-2.00   sec  2.55 MBytes  21.4 Mbits/sec    0    252 KBytes       
[  5]   2.00-3.00   sec  2.49 MBytes  20.9 Mbits/sec    0    355 KBytes       
[  5]   3.00-4.00   sec  2.42 MBytes  20.3 Mbits/sec    0    465 KBytes       
[  5]   4.00-5.00   sec  1.86 MBytes  15.6 Mbits/sec    1    339 KBytes       
[  5]   5.00-6.00   sec  1.86 MBytes  15.6 Mbits/sec    0    393 KBytes       
[  5]   6.00-7.00   sec  1.86 MBytes  15.6 Mbits/sec   39    298 KBytes       
[  5]   7.00-8.00   sec  1018 KBytes  8.34 Mbits/sec    1    216 KBytes       
[  5]   8.00-9.00   sec  2.86 MBytes  24.0 Mbits/sec    0    240 KBytes       
[  5]   9.00-10.00  sec  1.86 MBytes  15.6 Mbits/sec    0    255 KBytes       
[  5]  10.00-11.00  sec  1.86 MBytes  15.6 Mbits/sec    0    260 KBytes       
[  5]  11.00-12.00  sec  1.86 MBytes  15.6 Mbits/sec    0    260 KBytes       
[  5]  12.00-13.00  sec  1.86 MBytes  15.6 Mbits/sec    0    260 KBytes       
[  5]  13.00-14.00  sec  1.86 MBytes  15.6 Mbits/sec    0    263 KBytes       
[  5]  14.00-15.00  sec  1.86 MBytes  15.6 Mbits/sec    0    272 KBytes       
[  5]  15.00-16.00  sec  1.86 MBytes  15.6 Mbits/sec    0    288 KBytes       
[  5]  16.00-17.00  sec  1.86 MBytes  15.6 Mbits/sec    0    318 KBytes       
[  5]  17.00-18.00  sec  1.86 MBytes  15.6 Mbits/sec    0    361 KBytes       
[  5]  18.00-19.00  sec  1.86 MBytes  15.6 Mbits/sec    0    423 KBytes       
[  5]  19.00-20.00  sec  1.86 MBytes  15.6 Mbits/sec    0    506 KBytes       
[  5]  20.00-21.00  sec  2.17 MBytes  18.2 Mbits/sec    0    612 KBytes       
[  5]  21.00-22.00  sec  2.49 MBytes  20.9 Mbits/sec    0    742 KBytes       
[  5]  22.00-23.00  sec  1.25 MBytes  10.5 Mbits/sec    0    901 KBytes       
[  5]  23.00-24.00  sec  2.50 MBytes  21.0 Mbits/sec    0   1.06 MBytes       
[  5]  24.00-25.00  sec  1.25 MBytes  10.5 Mbits/sec    0   1.28 MBytes       
[  5]  25.00-26.00  sec  2.50 MBytes  21.0 Mbits/sec    0   1.52 MBytes       
[  5]  26.00-27.00  sec  1.25 MBytes  10.5 Mbits/sec    0   1.64 MBytes       
[  5]  27.00-28.00  sec  2.50 MBytes  21.0 Mbits/sec    0   1.64 MBytes       
[  5]  28.00-29.00  sec  1.25 MBytes  10.5 Mbits/sec    0   1.64 MBytes       
[  5]  29.00-30.00  sec  2.50 MBytes  21.0 Mbits/sec    0   1.64 MBytes       
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Retr
[  5]   0.00-30.00  sec  59.7 MBytes  16.7 Mbits/sec   41             sender
[  5]   0.00-30.59  sec  57.9 MBytes  15.9 Mbits/sec                  receiver

iperf Done.

Not quite as stable as a download, with peaks of 21-22Mbits but an average of 16Mbits with some retries. Not bad for a first try.

Given that the devices I’ve been working with lately are happy with a few hundred kilobits of internet access, a connection in the megabits is more than enough to get me started.

Conclusion

If you’ve read this far I hope my waffle helped get your setup working. Or at least helped save some of the Googling I’ve subjected myself to over the last few weeks.

Now with a basic setup in place there’s plenty left to try / refine.

The LimeSDR should be able to handle a 20Mhz wide channel (the configuration described above being for a 10Mhz channel). Which should increase its download throughput to around 75Mbits.

There’s also be the option of MIMO or carrier aggregation to explore as the LimeSDR has two channels. Allowing for a potential downlink of 150Mbps and uplink of 50Mbps in the future.

Was this article helpful?
YesNo

Updates

04/06/23

  • Updated method of setting EARFCN.
  • Updated launch examples based on PlutoSDR work.
, , , ,

27 responses to “Private LTE with LimeSDR and srsRAN – Part 4 (Config & Launch)”

  1. Hafithar avatar
    Hafithar

    I am using bladerf x40, my core network and eNB communicate but the UE doesn’t communicate with the network,any help,
    Thank you

    1. Phil Greenland avatar
      Phil Greenland

      Hi, I’ve not used a bladerf myself. If you’ve got access to a spectrum analyser, checking to see if the SDR is transmitting an LTE looking spectrum at the configured frequency would be my first port of call. I’ve found in the past with certain modems and SDRs the clock accuracy becomes an issue. For example I’ve a sierra wireless modem that generally connects to anything and a qualcomm based modem that’s very picky about the SDR being perfectly on frequency. Hope that might be of some help. Thanks, Phil

    2. Ray avatar
      Ray

      I would suggest checking the image loaded for the fpga in your bladeRF x40. The latest release of x40 has some issues but the 2019 releases seems to work for me.
      BladeRF has lots of tweaks to be done and I suggest the following ‘easy fix’:
      Operating system: DragonOS (Install or run live )
      BladeRF image: Download 2019 release and load it to your x40 fpga.

      While srsRAN has been configured in DragonOS, I also followed all steps here replacing soapy sdr specifics with bladeRF and everything worked as well.

      Note: the DragonOS has already installed nearly all the libraries needed.

  2. dave avatar
    dave

    fantastic job ,i just wanted to ask what are you planning on doing with this network ,and what do you think are the limits of SDR in LTE

    1. Phil Greenland avatar
      Phil Greenland

      Hi Dave,

      Thanks for that.

      One of the aims of using SDR technology for LTE networks / the development of future standards is the lower cost of entry, when compared to traditional special purpose hardware.

      An example use that’s often quoted is the development of picocells at a much reduced cost.

      I personally use my setup for testing LTE connected products I’m working on with my customers.
      Testing with public networks is important but the LTE simulator provides me with a level of control I can’t get over public networks.

      I can for example manipulate the signal strength, generate errors and change the network availability at will. Ensuring the customer’s modem (and quite often my code controlling it) takes the appropriate actions, in an attempt to get the device back online.

      I can interact with the network at different levels. For example simulating failures in services such as DNS, NTP or TCP/IP. Ensuring the device detects this and behaves (typically retries) appropriately. All without modifying the code on the device.

      There’s also, to a lesser extent reduced costs. I can push megabytes or gigabytes of data from a device to a local server or the internet without the expense of a cellular plan.

      Hope that helps,

      Regards,

      Phil

      1. Dm. Goreslavich avatar
        Dm. Goreslavich

        Don’t forget that ADALM-PLUTO is much cheaper than other SDRs. Except for Pluto+. Pluto+ is even cheaper!

        1.  avatar
          Anonymous

          but Pluto doesn’t have hardware timestamp so it cant work with LTE right ?

          1. Phil Greenland avatar
            Phil Greenland

            Not out of the box, but it can be added 🙂 – see Private LTE with Analog ADALM-PLUTO

  3. dave avatar
    dave

    Hello again ,so in the past 3 weeks I’ve been looking to create my own private LTE network and my initial thought was to do it with a SDR as its surely the cheapest way but after research i found that a SDR setup with a external clock and a duplexer using a limeSDR is same price maybe even more expensive than using a baicells eNB (nova 227 pbs2120) both are around 700$ and then suing a EPC of your choice (magma core or srsRAN or OpenAirInterface ).
    Did you consider other option when doing this or at the time this was the cheapest and most efficient way

    1. Phil Greenland avatar
      Phil Greenland

      I pretty much went straight down the SDR route, having seen a similar setup used by a previous employer. My thought process also covered the migration to 5G. srsRAN is now offering a 5G stack. Last time I checked this only supported the Ettus Research SDRs but should be capable of supporting others in time. Having a quick look at the Nova 227, it’s using TDD LTE rather than the more widely used FDD LTE. I doubt that would be much of an issue with most devices, aside from slightly reduced performance. For my current setup, given that its running over cables I’ve been using a power splitter rather than a diplexer to combine the TX and RX channels. The cheapest solution I could find and it’s not band specific, again not that thats too much of an issue.

      1. dave avatar
        dave

        i see , it would be exciting if you do a 5g network using lime
        well thanks for the response and once again great work on this and the Pluto one as well ,saw that and it was fantastic as well
        keep up the great work

  4. Thatguy avatar
    Thatguy

    Did you ever have reliability issues with the LimeSDR? Seeing a few complaints that the LimeSDR Mini only stays up for about 20 minutes and dies.

    1. Phil Greenland avatar
      Phil Greenland

      Hey,

      I may have been responsible for at least one of the 20 minute problem posts on the MyraidRF forum.

      Using the LimeSDR and LimeSDR Mini with srsRAN via SoapySDR I’ve not had any issues.

      I was using a mini in the setup I describe here for 8+hours last week without issue.

      I had issues in the past using srsRAN with the interface created by the LimeSDR team, which makes use of LimeSuite directly (dropping the need for the SoapySDR library). Weirdly it worked perfect with the original LimeSDR but would fail fairly consistently at the 20 minute mark with the mini. I’ve not revisited yet but will update my original forum post when I do.

      Hope that helps!

      Thanks,

      Phil

      1. thatguy avatar
        thatguy

        Appreciate the feedback! I thought I had something for a moment, but yours also has “driver=lime”.

        What I’ve been unable to figure out is how you managed to get the reference clock to 30.72 MHz, mine is stuck at 40 MHz and ignores the master_clock_rate argument. As far as I can tell, I’m matching your configuration. Are you willing to share the files, please?

        The other fun challenge I’m having is that when the LimeSDR stops working and I press Control-C or try to end it, it locks up the computer. Any thoughts on anything would be very appreciated.

        1. Phil Greenland avatar
          Phil Greenland

          The clock issue is almost certainly a red herring.
          The original LimeSDR-USB I was using for at the time of writing the blog post uses a Rakon RPT7050A TCXO @30.72MHz as its master clock, see: https://wiki.myriadrf.org/LimeSDR-USB
          The new LimeSDR-Mini v2 uses a Rakon E7355LF @40 MHz, see: https://limesdr-mini.myriadrf.org/documentation/board-description

          The drivers should calculate the correct clock dividers for the board you’re running.

          I’ve very occasionally seen the closing issue. The srsENB binary has a timeout mechanism, which after something like 5 seconds kills the stuck thread. That’s previously always got me back to the terminal. I don’t recall ever totally locking up the machine.

          Happy to zip up the config files I’ve been using. I’ll fire up the machine with an LimeSDR Mini v2 later to double check they’re all good then drop a link to them here.

        2. Phil Greenland avatar
          Phil Greenland

          Hey,

          I’ve tested and zipped the config files for my setup running with a LimeSDR Mini v2.0. I’ve also included the enb output when its starting and a device is connected. A test mobile device has been connected through it for the last 30+mins now. See how you get on with this: https://www.quantulum.co.uk/wp-content/uploads/2024/02/srsran_limesdr_mini_config.zip

          Thanks,

          Phil

          1. thatguy avatar
            thatguy

            Really appreciate it! I am noticing even with your configuration, it’s complaining about the following: “Detected LimeSDR. Consider using LTE rates for better RF performance. Either compile with ‘-DUSE_LTE_RATES=True’ or start srsENB/srsUE with ‘–expert.lte_sample_rates=true’.

            I’ve tried the latter, but it seems to complain either way. You seem to have different drivers than I do, as I see your active RF Plugins are different than mine. My setup is a bit more different than yours, I’ll send a message about it.

          2. Phil Greenland avatar
            Phil Greenland

            The supported RF drivers line Supported RF device list: soapy limesdr file likely only contains soapy and file in your setup. The limesdr here is a bit of an experimental driver, which interfaces with LimeMicrosystems driver directly, negating the need for soapy. However on the LimeSDR mini it was causing the network fails after almost exactly 20 minutes issue.

            In terms of the LTE samples warning. I’ve built my setup providing -DUSE_LTE_RATES=True when running cmake. I think I have tried the enb command line option before though (from memory) and it did seem to work.

            Drop me an example output from the enb app if you like, happy to take a look and see if I can spot anything obvious. You should be able to paste a relatively large file into my contact page. Or drop me an email using my name at my domain.

            Thanks,

            Phil

  5. pysim (s)hell avatar
    pysim (s)hell

    Anyone tried successfully this fairly old fork of srsRAN which is supposed to send SIB12 (CMAS alert) even before phone is authenticated to network?

    https://github.com/aluminiumi/srsLTE-plus-sib12

    I’ve tried both without proper SIM keys and then also with Sysmocom/Gialersim cards, I can see in the PCAP files ENB does send the SIB12 blocks.

    Also via QCSuper I verified the phone sees the SIB12 blocks, but there seems to be something missing, like PCCH Paging. But even after spending lot of time searching what exact fix is needed, no phone displays the alert (likely due to some packet like paging missing).

  6. Jaspreet Kaur avatar
    Jaspreet Kaur

    Dear,
    I have successfully established the ORAN environment using USRPs and can get access to the internet as well. But when I am trying to use the SIM card in the dongle, the EPC authenticates the SIM card IMSI and also assigns an IP address to the SIM card, but I am accessing the internet on the SIM card in the dongle. Can you suggest something to fix it?

    1. Phil Greenland avatar
      Phil Greenland

      Hi Jaspreet,

      Assuming you’re using srsRAN 4G, if you have an IP address assigned, you may be experiencing a routing issue.

      See: https://docs.srsran.com/projects/4g/en/rfsoc/usermanuals/source/srsepc/source/2_epc_getstarted.html#connecting-ues-to-the-internet

      Thanks,

      Phil

  7. Matthew Hilts avatar
    Matthew Hilts

    Hey Phil, great stuff here.

    I have an srsRAN 4G setup going in my lab right now, and I am testing out the UL and DL performance using iperf3. DL is as expected, but the UL is quite poor; sometimes it works okay and other times it varies wildly with low throughput.

    Have you encountered anything like this with iperf3 in the uplink? I have a hunch it might be something due to interactions between iperf and the eNB UL allocation process. Let me know if you have seen anything similar!

    1. Phil Greenland avatar
      Phil Greenland

      Hey Matthew,

      Thanks! I’ve not seen too many upload issues with the LimeSDR, although haven’t really been pushing it lately past my original testing.

      What sort of performance are you seeing? / What’s your current network configuration in terms of PRBs?

      Thanks,

      Phil

      1. Matthew Hilts avatar
        Matthew Hilts

        Hey Phil,

        I am running with 25 PRBs, so expecting ~18Mbps on the downlink and ~12Mbps on the uplink.

        The actual performance I am getting is 18Mbps on the downlink and ~5Mbps on the uplink. Also, when I use IPERF with UDP, I get the correct uplink rate, just with a steady 2% error rate. This is making me think that maybe the link just has enough errors that the IPERF TCP is not increasing the rate? SNR is very high so might be a buffer overflow or timing issue.

        1. Phil Greenland avatar
          Phil Greenland

          Gotcha, I remember seeing this exact sort of behaviour myself with a single connection, although that was with a different SDR. There are iperf options for multiple TCP streams, which may allow for better utilisation of the link. With the congestion control instances fighting for bandwidth a little.

          Phil

          1. Matthew Hilts avatar
            Matthew Hilts

            It’s so good to hear you’ve seen this too! And thanks for your time, I just want to confirm what you’re saying:

            Issue: Iperf3 gets expected uplink rate with UDP mode, but not with TCP mode.

            Possible solutions: Try iperf with multiple TCP streams, and dig into the congestion control.

            If you have time, could you provide a little more info on your thoughts on congestion control, and any other relevant details? Thanks a ton!

          2. Phil Greenland avatar
            Phil Greenland

            Thats a neat summary.

            When it comes to congestion control, the default used by Linux is the cubic I believe, which is pretty good in most situations, hence it being the default. I’ve expected with Google’s BBR which is supported by Linux, typically available as a module depending on your distribution.

            Current available algorithms can be found with:

            cat /proc/sys/net/ipv4/tcp_available_congestion_control

            Current kernel config options for congestion control something like:

            grep TCP_CONG /boot/config-$(uname -r)

            You can for example load BBR with:

            sudo modprobe tcp_bbr

            Which will then cause it to become available in the tcp_available_congestion_control list.

            Aside from the multiple streams in iperf, you’re able to select the congestion control algorithm you’d like to use, independently of the OS setting. With either -Z or --linux-congestion.

            As a high level overview / history this site was in my bookmarks: https://www.speedguide.net/articles/tcp-congestion-control-algorithms-comparison-7423

            While not directly related to your LTE work, cloudflare are often optimising for mobile, facing the same sort of issues due to packet loss. Pretty much all of their blog posts are excellent, this one while focused on HTTP/2 touches on BBR and congestion control: https://blog.cloudflare.com/http-2-prioritization-with-nginx/

            I looked into it in some detail a while ago for a customer using satellite links. Although we were looking to optimise for latency rather than loss in that case with the round trip to a satellite coming in at 600+ms.

            Thanks,

            Phil

Leave a Reply

Your email address will not be published. Required fields are marked *