标签: AI-RAN

  • AI-RAN : USRP X410 配置

    Spark 配置

    安装Sionna-rk

    git clone https://github.com/NVlabs/sionna-rk.git
    cd sionna-rk

    安装UHD(Orin平台请勿使用)

    sudo apt install -y \
        autoconf automake build-essential ccache cmake cpufrequtils \
        doxygen ethtool g++ git inetutils-tools libboost-all-dev \
        libncurses6 libncurses-dev libusb-1.0-0 libusb-1.0-0-dev \
        libusb-dev python3-dev python3-mako python3-numpy python3-requests \
        python3-scipy python3-setuptools python3-ruamel.yaml ninja-build
    
    git clone https://github.com/EttusResearch/uhd.git
    
    cd host
    mkdir build && cd build
    cmake -DCMAKE_POLICY_DEFAULT_CMD0167=NEW -GNinja ..
    ninja
    ninja test
    sudo ninja install

    需要建立Python的索引

    cd /usr/lib/python3/dist-packages
    sudo vim uhd.pth
    
    /usr/local/lib/python3.12/site-packages

    查看GPU信息

    nvidia-smi

    GPU信息显示如下:

    Thu Nov 27 19:33:48 2025       
    +-----------------------------------------------------------------------------------------+
    | NVIDIA-SMI 580.95.05              Driver Version: 580.95.05      CUDA Version: 13.0     |
    +-----------------------------------------+------------------------+----------------------+
    | GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
    | Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
    |                                         |                        |               MIG M. |
    |=========================================+========================+======================|
    |   0  NVIDIA GB10                    On  |   0000000F:01:00.0  On |                  N/A |
    | N/A   42C    P0              5W /  N/A  | Not Supported          |      0%      Default |
    |                                         |                        |                  N/A |
    +-----------------------------------------+------------------------+----------------------+
    

    安装 cupy

    pip install cupy-cuda13x --break-system-packages

    如果要让Spark的网口速率固定为10G,请执行指令

    sudo ethtool -s enP2p1s0f0np0 speed 10000 duplex full autoneg off

    检查速率

    sudo ethtool enP2p1s0f0np0 | grep -i speed

    因为不存在速率自动协商的功能,所以需要手动配置好。

    USRP X410 配置

    当你第一次拿到USRP X410的时候,请按照下图所示的常规的连接方法进行接线。

    • 将Spark和USRP都连接到同一台路由器中。
    • USRP的QSFP28口1和Spark的网口进行连接,注意正反。

    在Spark中启动一下终端,然后在终端中运行

    uhd_find_devices

    会得到如下的输出

    airan1@spark-bbdd:~$ uhd_find_devices 
    [INFO] [UHD] linux; GNU C++ version 13.3.0; Boost_108300; UHD_4.9.0.0-75-g0f522267
    --------------------------------------------------
    -- UHD Device 0
    --------------------------------------------------
    Device Address:
        serial: XXXXXXX
        addr: 192.168.0.104
        claimed: False
        fpga: X4_200
        mgmt_addr: 192.168.0.104
        product: x410
        type: x4xx

    然后输入指令进入终端

    airan1@spark-bbdd:~$ ssh root@192.168.0.104
    The authenticity of host '192.168.0.104 (192.168.0.104)' can't be established.
    ED25519 key fingerprint is SHA256:tu6m3SUsvFbVQxtLQiGBU/5BuEO3QexkXl4BNUh+Ye0.
    This key is not known by any other names.
    Are you sure you want to continue connecting (yes/no/[fingerprint])? yes
    Warning: Permanently added '192.168.0.104' (XXXXXXX) to the list of known hosts.
    

    进入终端之后,也可以查看设备信息

    root@ni-x4xx-XXXXXXX:~# uhd_find_devices 
    [INFO] [UHD] linux; GNU C++ version 9.2.0; Boost_107100; UHD_4.1.0.7-0-g8f9508f7
    --------------------------------------------------
    -- UHD Device 0
    --------------------------------------------------
    Device Address:
        serial: XXXXXXX
        claimed: False
        fpga: X4_200
        mgmt_addr: 127.0.0.1
        product: x410
        type: x4xx

    完成之后,可以看到UHD得版本是4.1,可以来下载完整的镜像

    root@ni-x4xx-XXXXXXX:~# uhd_images_downloader

    下载完成之后,可以查看下载好的镜像

    root@ni-x4xx-XXXXXXX:~# ls /usr/share/uhd/images/

    在4.1版本中,是只能看到“X4_200”

    为了能在Spark中使用,那么就需要安装和Spark一样的UHD版本。

    更新文件系统

    root@ni-x4xx-XXXXXXX:~# usrp_update_fs -t master
    
    /usr/lib/python3.7/site-packages/usrp_mpm/rpc_server.py:24: MonkeyPatchWarning: Monkey-patching ssl after ssl has already been imported may lead to errors, including RecursionError on Python 3.6. It may also silently lead to incorrect behaviour on Python 3.7. Please monkey-patch earlier. See https://github.com/gevent/gevent/issues/1016. Modules that had direct imports (NOT patched): ['urllib3.contrib.pyopenssl (/usr/lib/python3.7/site-packages/urllib3/contrib/pyopenssl.py)']. 
      monkey.patch_all()
    Downloading manifest file from https://raw.githubusercontent.com/EttusResearch/uhd/master/images/manifest.txt...
    [INFO] Using base URL: https://files.ettus.com/binaries/cache/
    [INFO] Images destination: /usr/share/uhd/images
    [INFO] Using manifest file at location: /tmp/tmpsb7jmy6_/manifest.txt
    The file size for this target (438.3 MiB) exceeds the download limit (100.0 MiB). Continue downloading? [y/N]y
    459548 kB / 459548 kB (100%) x4xx_common_mender_default-v4.9.0.0.zip
    [INFO] Images download complete.
    INFO[0000] Loaded configuration file: /etc/mender/mender.conf 
    INFO[0000] Mender running on partition: /dev/mmcblk0p2  
    INFO[0000] Start updating from local image file: [/usr/share/uhd/images/usrp_x4xx_fs.mender] 
    Installing Artifact of size 466179072...
    INFO[0000] No public key was provided for authenticating the artifact 
    INFO[0000] Update Module path "/usr/share/mender/modules/v3" could not be opened (open /usr/share/mender/modules/v3: no such file or directory). Update modules will not be available 
    INFO[0000] Opening device "/dev/mmcblk0p3" for writing  
    INFO[0000] Native sector size of block device /dev/mmcblk0p3 is 512 bytes. Mender will write in chunks of 1048576 bytes 
    ................................   0% 1024 KiB
    ................................   0% 2048 KiB
    ................................   0% 3072 KiB
    ................................   0% 4096 KiB
    ................................   1% 5120 KiB
    ................................   1% 6144 KiB
    ................................   1% 7168 KiB
    ................................   1% 8192 KiB
    ................................   2% 9216 KiB
    ................................   X
    ................................   X
    ................................  97% 444416 KiB
    ................................  97% 445440 KiB
    ................................  98% 446464 KiB
    ................................  98% 447488 KiB
    ................................  98% 448512 KiB
    ................................  98% 449536 KiB
    ................................  98% 450560 KiB
    ................................  99% 451584 KiB
    ................................  99% 452608 KiB
    ................................  99% 453632 KiB
    ................................  99% 454656 KiB
    ..................INFO[0153] All bytes were successfully written to the new partition 
    INFO[0153] The optimized block-device writer wrote a total of 3769 frames, where 1516 frames did need to be rewritten (i.e., skipped) 
    INFO[0155] Wrote 3951034368/3951034368 bytes to the inactive partition 
                   100% 455253 KiB
    INFO[0155] Enabling partition with new image installed to be a boot candidate: 3 
    Use -commit to update, or -rollback to roll back the update.
    At least one payload requested a reboot of the device it updated.
    Applied image. After reboot, check if everything works, and then run the command '$ mender -commit' to confirm (otherwise, this update will be undone).
    Note: Any data stored in this partition will be not accessible after reboot.
    Reboot now? [Yn] y
    Will reboot now. Hit Ctrl-C before the countdown expires to cancel.
    Rebooting in 3...
    2...
    1...
    Connection to 192.168.0.104 closed by remote host.
    Connection to 192.168.0.104 closed.

    也可以指定UHD版本

    usrp_update_fs -t v4.7.0.0

    清除ssh的垃圾信息

    ssh-keygen -f '/home/airan1/.ssh/known_hosts' -R '192.168.0.104'

    重新ssh登陆X410

    root@ni-x4xx-XXXXXXX:~# uhd_find_devices 
    [INFO] [UHD] linux; GNU C++ version 11.5.0; Boost_107800; UHD_4.9.0.0-0-g006d7f76
    --------------------------------------------------
    -- UHD Device 0
    --------------------------------------------------
    Device Address:
        serial: XXXXXXX
        claimed: False
        fpga: X4_200
        mgmt_addr: 127.0.0.1
        name: ni-x4xx-34E40F5
        product: x410
        type: x4xx

    查看对应的固件信息

    root@ni-x4xx-XXXXXXX:~# ls /usr/share/uhd/images/
    inventory.json		       usrp_x410_fpga_UC_200.dts.md5   usrp_x440_fpga_CG_1600.dts      usrp_x440_fpga_X4_1600.bit.md5  usrp_x440_fpga_X4_400.bit
    usrp_x410_fpga_CG_400.bit      usrp_x410_fpga_UC_200.rpt       usrp_x440_fpga_CG_1600.dts.md5  usrp_x440_fpga_X4_1600.dts      usrp_x440_fpga_X4_400.bit.md5
    usrp_x410_fpga_CG_400.bit.md5  usrp_x410_fpga_X4_200.bit       usrp_x440_fpga_CG_1600.rpt      usrp_x440_fpga_X4_1600.dts.md5  usrp_x440_fpga_X4_400.dts
    usrp_x410_fpga_CG_400.dts      usrp_x410_fpga_X4_200.bit.md5   usrp_x440_fpga_CG_400.bit       usrp_x440_fpga_X4_1600.rpt      usrp_x440_fpga_X4_400.dts.md5
    usrp_x410_fpga_CG_400.dts.md5  usrp_x410_fpga_X4_200.dts       usrp_x440_fpga_CG_400.bit.md5   usrp_x440_fpga_X4_200.bit       usrp_x440_fpga_X4_400.rpt
    usrp_x410_fpga_CG_400.rpt      usrp_x410_fpga_X4_200.dts.md5   usrp_x440_fpga_CG_400.dts       usrp_x440_fpga_X4_200.bit.md5
    usrp_x410_fpga_UC_200.bit      usrp_x410_fpga_X4_200.rpt       usrp_x440_fpga_CG_400.dts.md5   usrp_x440_fpga_X4_200.dts
    usrp_x410_fpga_UC_200.bit.md5  usrp_x440_fpga_CG_1600.bit      usrp_x440_fpga_CG_400.rpt       usrp_x440_fpga_X4_200.dts.md5
    usrp_x410_fpga_UC_200.dts      usrp_x440_fpga_CG_1600.bit.md5  usrp_x440_fpga_X4_1600.bit      usrp_x440_fpga_X4_200.rpt

    固件类型

    FPGA Image FlavorNumber of
    Channels
    Bandwidth
    per Channel
    QSFP28 Port 0 InterfaceQSFP28 Port 1 InterfaceDDC/DUCDRAM
    X4_2004 (2 per ZBX)200 MHz4x 10 GbE (All Lanes)UnusedYesYes (4 GiB, 4-Ch Replay)
    UC_2004 (2 per ZBX)200 MHzUnused100 GbEYesYes (4 GiB, 4-Ch Replay)
    CG_4004 (2 per ZBX)400 MHz100 GbE100 GbENoNo

    烧写支持100G的固件

    root@ni-x4xx-XXXXXXX:~# uhd_image_loader --args "type=x4xx,mgmt_addr=127.0.0.1,fpga=CG_400"
    
    [INFO] [UHD] linux; GNU C++ version 11.5.0; Boost_107800; UHD_4.9.0.0-0-g006d7f76
    [INFO] [MPMD] Initializing 1 device(s) in parallel with args: mgmt_addr=127.0.0.1,type=x4xx,product=x410,serial=XXXXXXX,name=ni-x4xx-XXXXXXX,fpga=X4_200,claimed=False,skip_init=1
    [INFO] [MPM.main] Launching USRP/MPM, version: 4.9.0.0-g006d7f76
    [INFO] [MPM.main] Spawning RPC process...
    [INFO] [MPM.PeriphManager] Device serial number: XXXXXXX
    [INFO] [MPM.PeriphManager.ClkMgr] Using Clock Configuration:
    DB0: Master Clock Rate: 245.76 MSps @Converter Rate 2.94912 GHz
    DB1: Master Clock Rate: 245.76 MSps @Converter Rate 2.94912 GHz
    [INFO] [MPM.PeriphManager] Initialized 2 daughterboard(s).
    [INFO] [MPM.PeriphManager] init() called with device args `boot_init=True,clock_source=internal,time_source=internal,initializing=True'.
    [INFO] [MPM.RPCServer] RPC server ready!
    [INFO] [MPM.RPCServer] Spawning watchdog task...
    [INFO] [MPMD] Claimed device without full initialization.
    [INFO] [MPMD IMAGE LOADER] Starting update. This may take a while.
    [INFO] [MPM.PeriphManager] Installing component `fpga'
    [INFO] [MPM.PeriphManager] Installing component `dts'
    [INFO] [MPM.RPCServer] Resetting peripheral manager.
    [INFO] [MPM.PeriphManager] Device serial number: XXXXXXX
    [INFO] [MPM.PeriphManager.ClkMgr] Using Clock Configuration:
    DB0: Master Clock Rate: 491.52 MSps @Converter Rate 2.94912 GHz
    DB1: Master Clock Rate: 491.52 MSps @Converter Rate 2.94912 GHz
    [INFO] [MPM.PeriphManager] Initialized 2 daughterboard(s).
    [INFO] [MPM.PeriphManager] init() called with device args `boot_init=True,clock_source=internal,time_source=internal,initializing=True'.
    [INFO] [MPMD IMAGE LOADER] Update component function succeeded.
    [INFO] [MPMD] Initializing 1 device(s) in parallel with args: mgmt_addr=127.0.0.1,type=x4xx,product=x410,serial=XXXXXXX,name=ni-x4xx-34E40F5,fpga=CG_400,claimed=False,find_all=1
    [INFO] [MPM.PeriphManager] init() called with device args `find_all=1,fpga=CG_400,mgmt_addr=127.0.0.1,name=ni-x4xx-XXXXXXX,product=x410,clock_source=internal,time_source=internal,initializing=True'.
    [INFO] [0/Radio#0] Clocking reconfigured, running ADC Self Cal on DB0...
    [INFO] [0/Radio#0] Calibrating channel 0...
    [INFO] [0/Radio#0] Calibrating channel 1...
    [INFO] [0/Radio#0] Calibrated 2 channels.
    [INFO] [0/Radio#1] Clocking reconfigured, running ADC Self Cal on DB1...
    [INFO] [0/Radio#1] Calibrating channel 2...
    [INFO] [0/Radio#1] Calibrating channel 3...
    [INFO] [0/Radio#1] Calibrated 2 channels.

    再次查看固件的版本

    root@ni-x4xx-XXXXXXX:~# uhd_find_devices 
    [INFO] [UHD] linux; GNU C++ version 11.5.0; Boost_107800; UHD_4.9.0.0-0-g006d7f76
    --------------------------------------------------
    -- UHD Device 0
    --------------------------------------------------
    Device Address:
        serial: XXXXXXX
        claimed: False
        fpga: CG_400
        mgmt_addr: 127.0.0.1
        name: ni-x4xx-XXXXXXX
        product: x410
        type: x4xx

    重启之后,是需要确认固件的,请执行该指令

    root@ni-x4xx-XXXXXXX:~# mender commit
    INFO[0000] Loaded configuration file: /etc/mender/mender.conf 
    INFO[0000] 'UpdateControlMapExpirationTimeSeconds' is not set in the Mender configuration file. Falling back to the default of 2*UpdatePollIntervalSeconds 
    INFO[0000] 'UpdateControlMapBootExpirationTimeSeconds' is not set in the Mender configuration file. Falling back to the default of 600 seconds 
    INFO[0000] Mender running on partition: /dev/mmcblk0p3  
    Committing Artifact...
    INFO[0000] Executing script: ArtifactCommit_Enter_10_migrate-configuration 
    INFO[0000] Committing update 

    在Spark中可以开启100G网口通道

    配置100G网口的IP地址

    注意IP地址:

    Interface NameDescriptionDefault ConfigurationConfiguration FileEx.: X4_xxx FPGA imageEx.: CG_xxx FPGA image
    eth0RJ45DHCPeth0.networkDHCPDHCP
    int0Internal169.254.0.1/24int0.network169.254.0.1/24169.254.0.1/24
    sfp0QSFP28 0 (4-lane interface or lane 0)192.168.10.2/24sfp0.network192.168.10.2/24192.168.10.2/24
    sfp0_1QSFP28 0 (lane 1)192.168.11.2/24sfp0_1.network192.168.11.2/24N/A
    sfp0_2QSFP28 0 (lane 2)192.168.12.2/24sfp0_2.network192.168.12.2/24N/A
    sfp0_3QSFP28 0 (lane 3)192.168.13.2/24sfp0_3.network192.168.13.2/24N/A
    sfp1QSFP28 1 (4-lane interface or lane 0)192.168.20.2/24sfp1.networkN/C192.168.20.2/24
    sfp1_1QSFP28 1 (lane 1)192.168.21.2/24sfp1_1.networkN/CN/A
    sfp1_2QSFP28 1 (lane 2)192.168.22.2/24sfp1_2.networkN/CN/A
    sfp1_3QSFP28 1 (lane 3)192.168.23.2/24sfp1_3.networkN/CN/A

    切换到Spark中,在Spark的终端中运行

    airan1@spark-bbdd:~$ uhd_find_devices 
    [INFO] [UHD] linux; GNU C++ version 13.3.0; Boost_108300; UHD_4.9.0.0-75-g0f522267
    --------------------------------------------------
    -- UHD Device 0
    --------------------------------------------------
    Device Address:
        serial: XXXXXXX
        addr: 192.168.10.2
        claimed: False
        fpga: CG_400
        mgmt_addr: 192.168.0.104
        mgmt_addr: 192.168.10.2
        name: ni-x4xx-XXXXXXX
        product: x410
        type: x4xx

    在输出中是可以看到已经显示出100G网口的IP信息了。

    安装一下Gnuradio

    sudo apt install gnuradio

    运行基础的频谱软件

    uhd_fft -f 2450M -s 500M

    运行在cpu的软件,会存在一些丢包情况,不够连续。

    系统配置

    方法一,修改缓冲区的大小

    sudo sysctl -w net.core.rmem_max=250000000
    sudo sysctl -w net.core.wmem_max=250000000

    查看修改之后的数值

    sysctl net.core.rmem_max
    sysctl net.core.wmem_max

    方法二,持久化该配置

    sudo nano /etc/sysctl.d/uhd.conf

    写入如下的内容

    net.core.rmem_max = 250000000
    net.core.wmem_max = 250000000

    保存后执行:

    sudo sysctl --system

    开启SCTP内核

    echo sctp | sudo tee /etc/modules-load.d/sctp.conf

    Docker增加信息

    # Add Docker's official GPG key:
    sudo install -m 0755 -d /etc/apt/keyrings
    sudo curl -fsSL https://download.docker.com/linux/ubuntu/gpg -o /etc/apt/keyrings/docker.asc
    sudo chmod a+r /etc/apt/keyrings/docker.asc
    
    # Add the repository to Apt sources:
    echo \
      "deb [arch=$(dpkg --print-architecture) signed-by=/etc/apt/keyrings/docker.asc] https://download.docker.com/linux/ubuntu \
      $(. /etc/os-release && echo "${UBUNTU_CODENAME:-$VERSION_CODENAME}") stable" | \
      sudo tee /etc/apt/sources.list.d/docker.list > /dev/null
    
    sudo nvidia-ctk runtime configure --runtime=docker
    sudo systemctl restart docker
    
    sudo usermod -aG docker $USER

    加载镜像

    docker load -i oai-amf-v2.1.0-1.2.tar
    docker load -i oai-ausf-v2.1.0-1.2.tar
    docker load -i oai-gnb-cuda.tar
    docker load -i oai-lmf-v2.1.0-1.2.tar
    docker load -i oai-nrf-v2.1.0-1.2.tar
    docker load -i oai-nr-ue-cuda.tar
    docker load -i oai-nssf-v2.1.0-1.2.tar
    docker load -i oai-smf-v2.1.0-1.2.tar
    docker load -i oai-udm-v2.1.0-1.2.tar
    docker load -i oai-udr-v2.1.0-1.2.tar
    docker load -i oai-upf-v2.1.0-1.2.tar
    docker load -i ran-base-cuda.tar
    docker load -i ran-build-cuda.tar
    docker load -i trf-gen-cn5g.tar
    docker load -i mysql-8-0.tar

    Python API使用

    创建USRP对象

    import uhd
    # X310、X410、Nxx系列单台设备
    usrp = uhd.usrp.MultiUSRP("addr=192.168.10.2") 
    # X310、X410、Nxx系列多台设备
    usrp = uhd.usrp.MultiUSRP("addr=192.168.10.2,addr=192.168.10.3") 
    # B210系列单台和多台
    usrp = uhd.usrp.MultiUSRP("serial=30BF56C,serial=30BF573") 

    配置参考时钟(10 MHz 和 PPS)

    usrp.set_clock_source("external")
    usrp.set_time_source("external")

    或者

    usrp.set_clock_source("gpsdo")
    usrp.set_time_source("gpsdo")

    接收信号

    samps = usrp.recv_num_samps(
        1e6, # Number of samples
        2.4e9, # Frequency in Hz
        1e6, # Sampling rate
        [0], # Receive on channel 0
        80, # 80 dB of RX gain
    )