We have a variety of commercial and open source LTE infrastructure at ORBIT, both on commercial and on generic hardware. We support various combinations of UE, eNB, and EPC.
Potential configurations are:
- OAI UE and OAI eNB with or without OAI EPC
- Commercial UE with OAI eNB with OAI EPC
- Commercial UE with Amarisoft eNB with Amarisoft EPC
- Commercial UE with OAI eNB with Amarisoft EPC
- Commercial UE with Airspan eNB and EPC
Nexus 5, Nexus 5x configuration notes
These settings are known to work for the above two phones, and the ORBIT MME.
- Network name: "Orbit Network"
- Short name: ORBIT
- pdn_type: ipv4
- apn: orbitA
- bearer: LTE
For these phones, to fully connect, the EPC must advertise an MTU of 1428, to account for GTP overhead on backhaul link with MTU 1500.
OAI eNB with Amarisoft EPC
- Amarisoft EPC running on bsc03out, IP 192.168.12.63, MTU 1564
- PLMN 00101
- OAI eNB running on sandbox1 node1-1, IP 192.168.12.82, MTU 1564
- MCC 001
- MNC 01
- Band 25
- Modulation FDD
- Bandwidth 5mhz
- Netgear 341u Dongle running on sandbox1 node1-2
- Commands run are as in the Amarisoft tutorial.
- Launch EPC on bsc03out (this should already be running, and there are REST commands to control it)
- On the netgear dongle, in the Minicom session, run at!reset
- Launch the eNB (this works around an issue where the eNB crashes on reconnect of existing session)
- If the eNB crashes, reset the UE prior to re-launching the eNB
- Client devices will connect on band 25 or band 41
- OAI band 41 FDD is unstable, and crashes
- Bandwidths greater than 5mhz crash the eNB on band41, and fail to connect to the UEs on band 25, desipite being officially supported
- Measured with Netgear 341u, band 25, 5mhz bandwidth
- iperf -s running on bsc03out, iperf -c 192.168.3.1 -r running on node1-2.sb1
- TCP: 2.85mbps upstream, 16.1 mbps downsteam
- UDP: 8.15 mbps upstream, with 0-1.7% loss, 1 packet received out of order
- UDP: Downstream fails to connect, or crashed eNB
- Error messages seen
- [RRC][N][eNB 0] Frame 627: received a DCCH 1 message on SRB 0 with Size 50 from UE b9b8 [RRC][I][FRAME 21158][eNB][MOD 00][RNTI b9b8] Received on DCCH 1 RRC_DCCH_DATA_IND [RRC][I][FRAME 21158][eNB][MOD 00][RNTI b9b8] [RAPROC] Logical Channel UL-DCCH, processing RRCConnectionSetupComplete from UE Segmentation fault (core dumped)
- [PHY][E][remove_harq_pid_from_freelist] /root/openairinterface5g/openair1/SCHED/phy_procedures_lte_eNb.c:154: critical error, get in touch with the authors
U[RRC][I][rrc_rx_tx] Removing UE b469 instance [RRC][W][rrc_eNB_free_UE] [eNB 0] Removing UE RNTI b469 [MAC][I][rrc_mac_remove_ue] Removing UE 0 from Primary CC_id 0 (rnti b469) [S1AP][W][s1ap_ue_context_release_req] Failed to find ue context associated with eNB ue s1ap id: 0 [S1AP][E][s1ap_eNB_task] Failed to find ue context associated with eNB ue s1ap id: 0 [RRC][I][rrc_rx_tx] Removing UE b469 instance [RRC][W][rrc_eNB_free_UE] [eNB 0] Removing UE RNTI b469 [MAC][I][rrc_mac_remove_ue] Removing UE -1 from Primary CC_id 0 (rnti b469) [S1AP][W][s1ap_ue_context_release_req] Failed to find ue context associated with eNB ue s1ap id: 0 [MAC][E][rrc_mac_remove_ue] error in mac_remove_ue(), could not find previous to -1 in UE_list, should never happen, Dumping UE list [S1AP][E][s1ap_eNB_task] Failed to find ue context associated with eNB ue s1ap id: 0 /root/openairinterface5g/targets/RT/USER/lte-softmodem.c exit_fun() Exiting OAI softmodem: mac_remove_ue: Problem in UE_list LL[RRC][I][rrc_eNB_remove_ue_context] [FRAME 00000][eNB][MOD 00][RNTI b469] Removed UE context
Amarisoft eNB and EPC
- Have simultaneous connection from netgear dongle on node1-2.sb1 and node2-5.outdoor
- To enable connectivity, specific or default routes should be set to eth2
- ip route del default
- ip route add dev eth2 default
OAI eNB and EPC
- All components functional, and communicate with each other.
- Commercial UE unable to register/associate. Repeatedly partially associated on bands 25 and 41.
- Error seen crashing MME component, complains that IMSI is too short, only 13 digits.
- IMSI is 15 digits, 001010000000001, but has two leading 0s. May be a bug in IMSI length parsing, test PLMN is only one with leading zeros, and eurecom using 20893 instead.
- Currently working images are msherman-oai-enb.ndz and msherman-oai-epc.ndz, with enb running on node1-1, and epc on either node1-2, or node2-1.