48 thoughts on “(obsolete) OpenAir, single machine Ubuntu 17.04, after major EPC update”

1. Hi Laurent –

   Thanks very much – much more usable setup having the entire network in one machine. I’ve gotten through the entire build process without a hitch, however now that I’m trying to run the lte-softmodem with a BladeRF, I am getting the following error:

   Entering ITTI signals handler
   got sync (eNB_thread_single)
   got sync (eNB_thread_synch)
   [DEBUG @ sync.c:485] Sample discontinuity detected @ buffer 86, message 1: Expected t=3512249, got t=3512282
   RX overrun (1475) is detected. t=3511997Got 252 samples. nsymps 7680
   [PHY][E]rx_rf: Asked for 7680 samples, got 252 from USRP
   /home/tom/openairinterface5g/targets/RT/USER/lte-softmodem.c exit_fun() Exiting OAI softmodem: problem receiving samples
   [DEBUG @ bladerf.c:352] Enable Module: RX – False
   [DEBUG @ bladerf.c:352] Enable Module: TX – False
   [UDP][W]Received TERMINATE_MESSAGE
   closing all tasks
   ready_tasks 0
   waiting for XFORMS thread
   stopping MODEM threads
   Killing eNB 0 processing threads
   Exiting eNB thread PRACH
   [DEBUG @ bladerf.c:352] Enable Module: RX – False

   I was just wondering if in making the appropriate patches, whether the lte-softmodem is now limited to only the USRP? I got the impression that the eNB thought it was connected to a USRP, despite being linked to a BladeRF. Everything else is working as expected.

   Thanks in advance –

   Tom

   1. Tom,

      We don’t have bladeRF (we have LimeSDR, USRP B2x0 and ExpressMIMO2).

      Nevertheless, the error you face is typical to all USB based boards: almost the same message can be obtained from USRP and LimeSDR.

      The eNB doesn’t read fast enough the Rx samples (or the board driver is buggy).

      I would say:
      1) verify the board is recognized as USB3.0
      (in the output of command “dmesg”, in bladeRF specific commands)
      2) try to reduce the modulated band
      in eNB configuration file, use N_RB_DL = 25; (for LTE over 5MHz band)
      3) Try a better computer
      Many people use Intel “U” processor, even expensive like i7-7200U, but it is only quite poor in term of performance as it is only 2 regular cores.
      Also, check the result of command “dmesg” after the error occurs: you may have thermal issues, usually traced like “PowerClamp”or similar messages

      Regards,
      Laurent

      1. Apologies – You were very right Laurent. USB 3.0 was not enabled in the BIOS.

         Sorry to keep bugging you, I am still having trouble after this to run the enodeb. I am using a quad core i7 machine at 2.7Ghz (a lenovo w530) which I believe to be powerful enough. I was using a dual core i5 at 2.6Ghz which was running lte-softmodem Release 10, however was a bit unstable – hence why I have upgraded to the quad-core i7 with lte-softmodem rel14. I have reduced the bandwidth to 1.4mhz (N_RB_DL=6), which seems to clear up the RX overrun issues, however I still keep running into some sync errors when trying to use the BladeRF as seen below.

         The error seems to be one of either : “Timeout (sent 0/7680)” (as below), OR, “RX : Expected ### samples, however got 0 from USRP”.

         Anshul Thankur helpfully noted that “sync_rx and sync_tx calls into bladeRF return 0 on success and a negative value on error. Contrast this to the way USRP code is built, where a send call returns the number of bytes/samples successfully sent (which is non-zero)”. Could this be a contributor to the problem somewhere in the code if the bladerf is treated as a USRP?

         Thanks and Warm Regards Laurent,

         Tom

         [DEBUG @ sync_worker.c:95] RX overrun @ buffer 122
         Sending sync to all threads
         got sync (eNB_thread_synch)
         got sync (eNB_thread_single)
         [DEBUG @ sync.c:704] Provided timestamp=26686718 is in past: current=26745598
         [PHY][E]TX : Timeout (sent 0/7680)
         /home/tom/openairinterface5g/targets/RT/USER/lte-softmodem.c exit_fun() Exiting OAI softmodem: problem transmitting samples
         [DEBUG @ bladerf.c:352] Enable Module: RX – False
         TYPE TO TERMINATE
         Entering ITTI signals handler
         [DEBUG @ bladerf.c:352] Enable Module: TX – False
         [UDP][W]Received TERMINATE_MESSAGE
         closing all tasks
         ready_tasks 0
         waiting for XFORMS thread
         stopping MODEM threads
         Killing eNB 0 processing threads
         Exiting eNB thread PRACH
         [DEBUG @ bladerf.c:352] Enable Module: RX – False

2. hi,laurent,
   The file “build_helper.patch” is empty, so the file no ptach!

   1. Ynnop,

      Thanks

      This is my mistake: take the file opencells-mods/build/tools/build_helper, it is the right one.
      To improve the flexibility, i computed patch files instead of delivering a new version only.
      But, I made the patch file wrong, so please take directly the new version of this file.

      I’ll fix the entire application note later, grouping several fixes together

      Laurent

3. Hello, it looks like the patch “opencells-mods/build/tools/build_helper.patch” in opencells-mods.tgz (18113f8ba3d97aaeafa9ae830bed6c6c39814978) is empty.

   1. This is a mistake, see the workaround sent on June 13th

      1. Thanks!

4. Hello Laurent,

   I saw that ubuntu 1704 provides stock uhd packages. Why not use it instead of compiling from source (which provides version 3.11 though).

   dpkg -l|grep uhd
   ii libgnuradio-uhd3.7.10:amd64 3.7.10.1-2 amd64 gnuradio universal hardware driver functions
   ii libuhd003:amd64 3.9.5-2 amd64 universal hardware driver for Ettus Research products – library
   ii soapysdr0.5-2-module-uhd:amd64 0.3.2-1 amd64 UHD device support for SoapySDR
   ii uhd-host 3.9.5-2 amd64 universal hardware driver for Ettus Research products – host apps

5. Hello,i have attempted your instructions on Ubuntu 17.04 ,but when i run ./build_hss,there is an error as fllows:
   root@zyl-ThinkStation-P310:/home/zyl/openair-cn/scripts# ./build_hss
   ./build_hss: 行 161: -DOPENAIRCN_DIR=/home/zyl/openair-cn: 没有那个文件或目录
   make: *** 没有指明目标并且找不到 makefile。 停止。
   cp: 无法获取’/home/zyl/openair-cn/build/hss/build/oai_hss’ 的文件状态(stat): 没有那个文件或目录
   i do not know how to resolve this problem,could you help me?Thank you very much

   1. Zyl,
      The message is in not in English, i can’t read it.
      A general advise is to install Linux with English language.
      Regards,
      Laurent

   2. Hello,I met the same problem. When I run ./build_hss -i ,the error is
      …
      0 upgraded, 0 newly installed, 0 to remove and 142 not upgraded.
      HSS not compiled, to compile it, re-run build_hss without -i option
      Then I run ./build_hss, the error is
      ./build_hss: line 161: -DOPENAIRCN_DIR=/home/wjj/Downloads/openairinterface5g/openair-cn: No such file or directory
      make: *** No targets specified and no makefile found. Stop.
      cp: cannot stat ‘/home/wjj/Downloads/openairinterface5g/openair-cn/build/hss/build/oai_hss’: No such file or directory

      1. Today I re-installed all the files, when I met step patch EPC, I used
         wget http://open-cells.com/d5138782a8739209ec5760865b1e53b0/opencells-mods-20170710.tgz
         tar xf opencells-mods-20170710.tgz
         git clone https://gitlab.eurecom.fr/oai/openair-cn.git
         cd openair-cn
         git checkout develop
         git apply ../opencells-mods/EPC.patch
         and succeed.
         Though I don’t know why.
         Thank you !

         1. Wjj,
            Thanks for sharing this good news
            Laurent

6. Hi Laurent,

   Thanks for the excellent post! I was wondering if it would be possible to see the complete contents of your enb.conf file? I have attempted to re-create this with the LimeSDR and everything compiled/patched/etc. just fine (I followed the steps on your Lime-specific post) but when I attempt to run lte_softmodem I get the following error:

   ==================================

   root@ubuntu:~/openairinterface5g# ./cmake_targets/lte_build_oai/build/lte-softmodem -d -O ./targets/PROJECTS/E-MBMS/enb.conf –rf-config-file ./targets/ARCH/LMSSDR/LimeSDR_above_1p8GHz_1v4.ini
   # /dev/cpu_dma_latency set to 0us
   log init done
   Running with XFORMS!
   WARNING: setting ‘tracking_area_code’ not found in configuration file

   Assertion (0) failed!
   In enb_config_init() /home/paul/openairinterface5g/openair2/ENB_APP/enb_config.c:797
   Failed to parse eNB configuration file ./targets/PROJECTS/E-MBMS/enb.conf, 0 th enb

   Exiting execution
   ============================

   The following is the contents of the enb.conf file I’m attempting to use:

   ============================

   Active_eNBs = ( “eNB_Eurecom_0”);

   eNBs =
   (
   {
   ////////// Identification parameters:
   eNB_ID = 347472;

   cell_type = “CELL_MACRO_ENB”;

   eNB_name = “eNB_Eurecom_0”;

   // Tracking area code, 0x0000 and 0xfffe are reserved values
   tracking_area_code = 1;

   mobile_country_code = 208;

   mobile_network_code = 92;

   ////////// Channel parameters:
   // Default Paging DRX of the eNB as defined in TS 36.304
   default_paging_drx = “PAGING_DRX_256”;

   ////////// Physical parameters:
   frame_type = “TDD”;
   prefix_type = “NORMAL”;
   eutra_band = 33;
   downlink_frequency = 1907.6e6;
   uplink_frequency_offset = 0.0;

   ////////// MME parameters:
   mme_ip_address = ( { ipv4 = “127.0.0.20”;
   ipv6 = “192:168:30::17”;
   active = “yes”;
   preference = “ipv4”;
   });

   NETWORK_INTERFACES :
   {
   ENB_INTERFACE_NAME_FOR_S1_MME = “lo”;
   ENB_IPV4_ADDRESS_FOR_S1_MME = “127.0.0.10/8”;

   ENB_INTERFACE_NAME_FOR_S1U = “lo”;
   ENB_IPV4_ADDRESS_FOR_S1U = “127.0.0.10/8”;
   ENB_PORT_FOR_S1U = 2152; # Spec 2152
   };
   }
   );

   ==============================
   I attempted to follow the instructions for modifying enb.conf, but this doesn’t seem to work. I also get the same error message when attempting to use the default enb.conf file. Any help/advice would be greatly appreciated! Thanks again for the great post!

   1. Hi Paul,

      With USRP B2x0, you can use the default file: targets/PROJECTS/GENERIC-LTE-EPC/CONF/enb.band7.tm1.50PRB.usrpb210.conf
      this is the one we used (or 100PRB one for max throughput)

      For Lime, there is config files also, at least it will fix your issue about file syntax.

      But, you may face later radio issues:
      – TDD: i don’t know anybody that run good TDD with Lime (even USRP has issues, see openair mailing list)
      – FDD: we didn’t push performance results because it was quite poor and not reproducible

      The Lime calibration was not working well, we helped to fix it in Tx, now it is not perfect but good enough and stable.
      On Tx, we have good EVM near 700MHz decreasing slowly to quite bad EVM at 3.5GHz.
      Band 7 (2.6GHz) is still acceptable: EVM below 4% (measured with a new R&S LTE analyzer)
      On Rx, it seems the calibration is still not very stable.

      Currently we are working on our framework to run OAI.
      So, we will re-try by end of July a Lime tuning session.

      You can get some details on: https://discourse.myriadrf.org/ and on OpenAir mailing list.

      We invested in LimeSDR, so we are eager it reach good quality.
      It is progressing, but still not good enough.

      Some measurements:
      https://open-cells.com/RS-results/RS-snapshots-20170523.tgz
      Since these measurements, Lime fixed issues, we re-tested: the bad I/Q figures (like in the file named “funny”) were fixed: Tx calibration is now stable.

      Laurent

7. Hi,

   I was following this tutorial and when i try to build the eNB i get the following message:

   joao@IDTS:~/IDTS/openairinterface5g$ ./cmake_targets/build_oai -I
   Will install external packages
   CMAKE_CMD=cmake ..
   RF HW set to None
   Flags for Deadline scheduler: False
   Flags for CPU Affinity: False
   2. Setting the OAI PATHS …
   OPENAIR_DIR = /home/joao/IDTS/openairinterface5g
   Installing packages
   ./cmake_targets/build_oai: line 457: check_install_oai_software: command not found

   Can you help me out please?

   1. Hi,
      I don’t know what you did.
      We missed to add unmodified files from June version.
      So, the file eNB build_helper was completely missing for the eNB in the patch file we made on July 6th.
      Whatever, the patch file of July 6th was not good and not fully tested.

      We retested today, from a fresh Ubuntu 17.04 installation to UE performance measurement.
      The patch file is re-built from this test.

      Regards,
      Laurent

8. Hi Laurent, thank you for the help, I was able to build the eNB already, although had to install freeDiameter for myseld(don’t know if necessary for eNB). But now i am trying to apply the patch to the EPC and it does not apply.

   joao@idts:~/idts/openair-cn$ git apply ../opencells-mods/EPC.patch
   error: patch failed: build/tools/build_helper:99
   error: build/tools/build_helper: patch does not apply

   I used the commit you provided in this tutorial. Did i do something wrong?

   Thank you for your attention.

   Cumprimentos,
   João Amaral

9. Hi Laurent,

   I believe that the commit hash for the EPC is not correct, i’ve downloaded the code from this commit and the patch is not appliable, whereas it is with the latest code, thank you.

   1. Joao,

      We’ve re-run several times.
      Could you be more specific on what is failing in this abstract sequence:

      wget http://open-cells.com/d5138782a8739209ec5760865b1e53b0/opencells-mods-20170710.tgz
      tar xf opencells-mods-20170710.tgz
      git clone https://gitlab.eurecom.fr/oai/openair-cn.git
      cd openair-cn
      git checkout develop
      git apply ../opencells-mods/EPC.patch

      I’ve re-re-run it right now, on latest commit of develop branch

      Regards,
      Laurent

10. hi,laurent,
    Thank you for your such detailed steps,and now I can run EPC and ENB correctly.But i met some problem for find signal with phone(support for 4G ).I write my card with:
    IMSI:208920100001100
    KI:6874736969202073796d4b2079650a73
    opc:504f20634f6320504f50206363500a4f

    In phone:
    apn:ltebox

    that’s all setting.
    When insert the card,my phone can not find the signal.
    Should i forget some necessary steps?
    thank you.

    best regards
    John

    1. Hi John,

       I don’t see any error in what you describe.

       Could check if the phone detects the beacon: in the phone setup, choose option “select manually the network”.
       When the phone finish the RF scanning you should see the network 208/92 is the list of detected networks.

       If yes, then check the OAI component traces: you should see the phone trying to access the network (message RACH in eNB, that should be followed by mme and HSS authentication process).

       Let’s see this trial, then next step
       Laurent

11. zb@ubuntu:~/openair-cn/scripts$ ./build_mme
    [sudo] password for zb:
    Build type is Debug
    Architecture is x86_64
    git found: /usr/bin/git
    NETTLE VERSION_INSTALLED = 3.3
    NETTLE_VERSION_MAJOR = 3
    NETTLE_VERSION_MINOR = 3
    mme compilation failed
    cp: cannot stat ‘/home/zb/openair-cn/build/mme/build/mme’: No such file or directory
    auth_request compilation failed
    cp: cannot stat ‘/home/zb/openair-cn/build/mme/build/auth_request’: No such file or directory

    1. Dear John,
       Your trace is very clear: mme compilation failed
       Thanks for being so accurate and polite
       Best regards,
       Laurent

12. Hi,

    Thanks for this detailed tutorial.

    I followed all the instruction in here but when i am trying to connect the UE i get the following error:
    from the Enodb:

    [S1AP][E][s1ap_eNB_task] Failed to find ue context associated with eNB ue s1ap id: 16398672

    from mme:

    | Current Status| Added since last display| Removed since last display |
    Connected eNBs| 1 | 0 | 0 |
    Attached UEs | 0 | 0 | 0 |
    Connected UEs | 0 | 1 | 1 |
    Default Bearers | 0 | 0 | 0 |
    S1-U Bearers | 0 | 0 | 0 |

    Any suggestion on what i did wrong?

    1. Dani,

       There is not enough data to make a accurate diagnostic.
       The most probable is the Radio link is poor, so the UE attaches, then loose link/re-attach/loose line …
       In this case, major bugs still exist in OAI that becomes confused like this: objects identifiers are not well managed (a lot of issues are related to race conditions inter-threads).

       If you send full logs, either on OAI mailing list or hereafter (a URL, please don’t copy thousands of lines in the blog)
       I will make a more accurate answer.

       Regards,
       Laurent

       1. Hi,

          I’ve met the same error with Dani just now, though I reinstall it for several times,the error still exist.I’m not able to find out whether it was caused by the codes or some steps I’ve missed when i was following your instructions.
          Could you please try it again when you are free, I’m really confused and upset now.thanks

          demo

    2. Hi.
       Have you resolved it?the error I met is the same with me ,if you have find a solution,could you teach me how to solve it.

       thanks

13. Hi!

    I’ve followed the tutorial but when I’m trying to connect with a UE (Huawei E3372 Dongle) I’m getting these errors:

    eNB:
    [S1AP][E][s1ap_eNB_task] Failed to find ue context associated with eNB ue s1ap id: 420141

    mme:
    INFO NAS-EM ir-cn/src/nas/emm/sap/emm_recv.c:0786 EMMAS-SAP – Received Authentication Failure message

    Any ideas what am i doing wrong?

    I’ve uploaded the complete logs to pastebin:
    https://pastebin.com/LZC0ApTs # hss
    https://pastebin.com/8FsFa8yz # mme
    https://pastebin.com/96UBviAs # eNB

    Best regards,
    Daniel

    1. Hello Daniel,
       Checking the first log: hss.log, i see you didn’t apply the full tutorial: the hss starts to recompile all OPc.
       We offer this tuto, nevertheless we can’t debug each user specific configuration.
       Regards,
       Laurent

       1. Hi laurent,

          I think the output was due to Daniel canceled the note on op in hss.conf, but whatever, I got the same result (Authentication failure :mac failure).
          I’m sure I’ve input the correct imsi,key and OPc into the sim card. The phone I use to attach someone has already tested, then I’m confused.

          thanks

          demo

       2. Hi!

          Thanks for the reply, yes I have accidentally uncommented the operator key line in the hss.conf just as demo said. I have solved this problem and now the hss doesn’t recompile the OPc values.

          Unfortunately, my problem still exists, I’m getting the same ‘Authentication failure’ message.

          The updated logs:
          https://pastebin.com/LZC0ApTs # hss
          https://pastebin.com/8FsFa8yz # mme
          https://pastebin.com/96UBviAs # eNB

          Best regards,
          Daniel

          1. Hi Daniel,

             I see in your trace: Entering nas_proc_ul_transfer_ind(), then emm_recv_authentication_failure()
             So, I believe the UE rejects the HSS authentication data.
             This occurs mainly because the OPc or the SQN is wrong.

             I don’t know which UICC/SIM you use, this could be tricky to debug (verify your OPc and SQN if you can).
             We decided to offer SIM cards for this purpose: we lost months to understand a small error like this one.

             Anyway, the radio attach is ok, IMSI is ok, the hss finds it in the DB, all the dialog is good.
             My understanding: the UE (probably the SIM) refuses data coming from the HSS.

             Regards,
             Laurent

             1. Hi!

                Thank you both for the detailed answers. It really seems like that my SIM OPc is wrong, I’m going to check that or try it with another SIM.

                Best regards,
                Daniel

          2. Hi Daniel,

             It might be caused by the sim card, I’m not sure.But Daniel has pointed it out clearly. I’ve bought the nano cards yesterday, maybe you can have a try as well.

             Regards,

             demo

14. Thanks for the amazing job.
    Yet, I just encounter some problem. Everything work fine before I try to connect Nexus 5 to the eNodeB. run.mme breaks down and the error logs are:

    Assertion (0) failed!
    In mme_app_handle_s6a_update_location_ans() /home/lrf/workspace/openair-cn/src/mme_app/mme_app_location.c:115
    “ULR/ULA procedure returned non success\n”
    Will Intentionaly raise SEGFAULT to be catched by GDB!
    Obtained 5 stack frames.
    mme(display_backtrace+0x28) [0x507251]
    mme(mme_app_handle_s6a_update_location_ans+0x283) [0x4faef6]
    mme(mme_app_thread+0x130) [0x4f4e2d]
    /lib/x86_64-linux-gnu/libpthread.so.0(+0x76ba) [0x7feea6b016ba]
    /lib/x86_64-linux-gnu/libc.so.6(clone+0x6d) [0x7feea50c13dd]
    ./run_mme: line 87: 2403 Segmentation fault (core dumped) $SUDO mme `echo $exe_arguments` 2>&1

    Do you have any idea what is going wrong?

    1. Oh.. I managed to solve the problem. I should have modified the IMSI ofpdn table in hss database. Still, nexus 5 cannot connect to the eNobeB. No error log is shown.

       1. Hi Ruofeng
          I got exactly the same error like you. i already put the ismi is the pdn table in the data base but the error persists. do you solve this problem ? do you know another way to solve this problem apart update the pdn table in the data base

          1. Hi,
             I don’t have nexus 5, so i can’t reproduce it.
             But, we’ve got a problem that looks similar with another Android phone.
             Could you try the patch in our post that updates this one (the August 22nd version)?
             Laurent

    2. Hi,

       Did you manage to connect Nexus 5? I’m facing the same problem.

       Thanks,

       Faouzi

15. I found that saying yes to “Do you want to install asn1c rev 1516 patched? : no” fixed my problem with the “cannot stat” issue that other users reported. I bash -x ./build_mme and saw error complaining about missing asn.1 compiler.

    1. Tim,
       Except if you have a specific use case, it would be better to switch to the new version of this post
       Laurent

       1. Thanks! Came here from a search for the stat error, didn’t see the August 22 post. Will use that now!
          -Tim

16. Hi,laurent
    first,thanks for your shareing.now i got a problem,i want to know what the address like MME_IPV4_ADDRESS_FOR_S11_MME = “127.0.11.1/8” in “mme.conf” for? Lots of people do not change the ip like this when they build mme/spgw. why we do not need to chang it. could you help me you? Thank you!

17. Leaf,
    It changes the MME source IP address over S11 (to the S-GW)
    If you want to observe: wireshark => open loopback interface => filter “udp.port==2123”
    You’ll see the IP source changes according to the parameter

    Is it useful to change it?
    No, as long as the IP packets can be routed by Linux kernel
    I change it for simple and clear configuration: “my” MME has one single IP addr, it uses it on all it’s interfaces

    Regards,
    Laurent

18. for Downloading the modifications

    download: patches extract the data:
    tar xf opencells-mods-20170710.tgz
    the error is :
    tar: opencells-mods-20170710.tgz: Cannot open: No such file or directory
    tar: Error is not recoverable: exiting now

    for downloading & Compiling the eNB on 17.04
    git apply opencells-mods/eNB.patch
    the error is : error: can’t open patch ‘opencells-mods/eNB.patch’: No such file or directory

    and Is the installing of SW packages depend on the commands before it? I got an error .

    Finally Thank you for your great articles which are more than useful.

    1. Ensaf,

       To extract files from a tar, you need to give the file path.
       Example: if you download the file in $HOME/Downloads (the default directory in firefox),
       then, supposing you are in $HOME (/home/ensaf), the command would be tar xf Downloads/opencells-mods-20170710.tgz

       Same problem with “git apply opencells-mods/eNB.patch”: you should have extract the directory “opencells-mods” from the tar file first, then provide the full path.

       We don’t think you will be successful with OpenAirInterface if you don’t know enough in Linux usage: you would have to learn the usual commands, the basic notion of files, directories, …

       Regards,
       Laurent

Comments are closed.