NI 射頻與無線平臺新產品介紹和應用分享_NI.pdf

《NI 射頻與無線平臺新產品介紹和應用分享_NI.pdf》由會員分享，可在線閱讀，更多相關《NI 射頻與無線平臺新產品介紹和應用分享_NI.pdf（46頁珍藏版）》請在三個皮匠報告上搜索。

1、NI射頻與無線平臺新產品介紹和應用分享NI RF&Wireless Product Growth ManagerHang CaoResearch Labon NI PlatformA Flexible Platform for Next Generation Wireless Research and PrototypingCPUGPPFPGADSPVCOPLLPLLD/AD/AA/DA/DVCORF Front End General Purpose RF Dual LOs Contiguous Frequency RangeHost processor Medium Access Cont

2、rol(MAC)Rx/Tx control ex.Host GPP,multi-core CPUStreaming DSPReal-time signal processor Physical Layer(PHY)ex FPGA,DSPBaseband ConvertersHost ConnectionDetermines Streaming Bandwidth Ex.Gigabit E-net,PCIeSoftware Defined Radio ArchitectureRxRadar PrototypingSpectrum Monitoring&PolicyRxSignal Intelli

3、genceTarget EmulationCommunication Protocol DesignSDR Technology Solves a Wide Range of ApplicationsAI for WirelessSensing&CommunicationsNon-Terrestrial NetworksOpen RANO-DUO-CUCNO-RUN321200 MHz BW6 GHz FcNI Ettus USRP X410400 MHz BW7.2 GHz FcStand AloneFPGA+Embedded ProcessorB200mini56 MHz BW6 GHz

4、FcUSRP 290156 MHz BW6 GHz FcHost based(USB)Low SWaPFrom Portable to High-performanceUSRP X310160 MHz BW6 GHz FcFlexRIO200 MHz BW4.4 GHz FcLarge FPGA SDRsUSRP E31X&E32056 MHz BW6 GHz FcDeployableUSRP RX310Pixus Technologies2nd Gen VST1 GHz BW6 GHz FcHigh Frequency/Wide BandwidthInstrument Grade+Calib

5、rationmmWave VST2 GHz BW54 GHz Fc3rd Gen VST2 GHz BW26.5 GHz FcWireless Research Design Prototyping&Deployment PortfolioTwo Platforms for RF CapabilitiesLow-SWAP Prototyping with USRP Open-Source Software CompatibleIntegration of RF with baseband and digitalSoftware migration path to tactical hardwa

6、reHigh-Performance Prototyping with PXIInstrument-grade hardwareHigher frequency and bandwidth optionAdvanced Applications|5G Test,Radar/EW Test,Communication Nav TestCalibrated,wideband vector signal analyzer:Frequency Range:30 MHz to 26.5 GHzUp to 2 GHz Instantaneous IQ BWProgrammable gain ranging

7、:+25 dBm max Independent or coherent operation with signal generatorSupports multi-channel synchronization and coherencyNI VST=Vector Signal Analyzer+Vector Signal Generator+Software Defined RadioCalibrated Wideband Signal Generator:Frequency Range:30 MHz to 26.5 GHzUp to 2 GHz Instantaneous IQ BW R

8、F Output Power:+20 dBmIntegrated Pulse Modulation,100 dB on/off ratiosIndependent or coherent operation with signal analyzerSupports multi-channel synchronization and coherencySoftware Defined Radio Capability/Common Expandable FPGAFull IBW Digital Streaming up to 4GHzNI Provided Expansion Applicati

9、ons(MEG,RPS,RTG)User Developed/Defined Expanded CapabilityNI Solution|NI Solution|Vector Signal Transceiver APXIe-7903 2-Slot High Speed Serial Coprocessor12 MiniSAS zHD Connectors48 RX/TX with 28.2 Gbps line rate External Reference Clock Input/OutputXilinx Virtex UltraScale+VU11P FPGA9,200+DSP slic

10、es20 GB of DRAM w/up to 25 GB/s write speed8 GPIO(mini-HDMI)PCI Express Gen 3x8341 Mb of Embedded Block and Ultra RAMData movement with Ethernet protocols100 GbE(previously 10 GbE was limit)Aurora 64b/66b NIs most powerful,highest speed,PXI FPGA coprocessor,built to maximize data movement and comput

11、ational High Speed Serial and Coprocessor PortfolioModel NamePXIe-6594PXIe-7902PXIe-7915PXIe-7903I/O8 RX/TX(MGTs)8 DIO24 RX/TX(MGTs)4 RX/TX(MGTs)8 DIO48 RX/TX(MGTs)Maximum Serial Data Rate(per channel)28 Gb/s12.5 Gb/s16.4 Gb/s28.2 Gb/sFPGAKintex UltraScale+KU15PVirtex-7 485TKintex UltraScale KU060Vi

12、rtex UltraScale+VU11PDynamic RAM 8 GB2 GB4 GB20 GBBlock RAM34.6 Mb37.1 Mb38.0 Mb341 MbDSP Slices1968280027609216PXI Backplane LinkPCIe Gen3 x8PCIe Gen2 x8PCIe Gen3 x8PCIe Gen3 RF/Digital Systems PerformanceWideband RF Recording 1 GHz IBW per channel100 GbE Stream To Disk40 GB/s(8 GHz IBW)360 TB=2.5

13、hours at max IBWMulti-Channel/FrequencySelect Partner/Customer Inline or Parallel ProcessingOffline AnalysisSystem Calibration&SyncTime Synchronization w/NI-TClkRecord Trigger OptionsSoftware,DigitalLO Power Calibration Wideband Multi-Channel System Equalization CalibrationMultichannel RF Record&Pla

14、yback SolutionControlData100 GbEEthernet controlOffline Analysis DataControl&Real-Time Analysis PXI ConfigurationPXIe-5841 SupportedSystem SoftwareRemote-access with gRPCSigMF metadata file formatEttus USRPB2XXEttus USRP E320Ettus USRPN310/N32XEttus USRP X310Ettus USRPX410Ettus USRPX440Frequency70 M

15、Hz 6 GHz70 MHz 6 GHz3 MHz-6 GHz(N32X)10 MHz-6 GHz(N310)*10MHz 6 GHz1MHz 7.2 GHz30MHz 4 GHz IFBandwidth56 MHz56 MHz200 MHz(N32X)100 MHz(N310)*160 MHz400 MHzUp to 1.6 GHzChannels2 Tx,2 Rx2 Tx,2 Rx2 Tx,2 Rx(N32X)4 Tx,4 Rx(N310)2 Tx,2 Rx*4 Rx(TwinRx)4 Tx,4 Rx8 Tx,8 RxArchitectureIntegratedIntegratedInte

16、grated*Configurable w/DaughterboardsIntegratedIntegratedCommunicationUSB10 GbE10 GbE or PCIe10 GbE or PCIe100/10 GbE or PCIe100 GbESynchronization2x2 MIMO2x2 MIMOUp to 128x128(N32X)Full Phase Synchronization*2x2 MIMO4x4 MIMO8x8 MIMOSW SupportGNU Radio,C+,Python,MATLAB,LabVIEWGNU Radio,C+,Python,MATL

17、AB,RFNoCGNU Radio,C+,Python,MATLAB,RFNoCGNU Radio,C+,Python,MATLAB,RFNoC,LabVIEW,LabVIEW FPGAGNU Radio,C+,Python,MATLAB,RFNoC,LabVIEW,LabVIEW FPGAGNU Radio,C+,Python,RFNoCKey FeaturesLow SWAP-C,Highly portableLow SWAP,Embedded Deployable,StandaloneStand Alone,Wide bandwidth,Multi-Channel Sync Ready(

18、N32X)*Configurable RF Front End,Programable FPGARFSoC based FPGA,5G Ready,Wide bandwidth,Multi-ChannelRFSoC based direct sampling,phase coherency,wide bandwidthUSRP Product Portfolio OverviewNI Ettus USRP X410 Product OverviewFrequency Range:1 MHz -7.2 GHzSignal Bandwidth:400 MHzReceive Channels:4XT

19、ransmit Channels:4XMax TX Power:up to 22 dBm1Max RX Power:0 dBmXilinx Zynq UltraScale+RFSOC Built-in quad core ARM processorOnboard IP:SD-FEC,DDC,DUCInterface options:dual QSFP28,PCIe Gen 3 x8Synchronization:10 MHz/PPS,GPSDO optionSoftware:Open source(GNU Radio,RFNoC,UHD)Matlab NI-USRP,LabVIEW FPGAR

20、F CapabilitiesDigital Capabilities1 see specification for detailsNI Supports Many Tool Flows13System ModelC/C+GPP DesignFPGA DesignRFNoC via GRCLabVIEW FPGADeploySDR HardwareSDR HardwareSDR HardwareWireless TestbenchWireless TestbenchWireless Testbench#!/usr/bin/env pythonfrom gnuradio import grfrom

21、 gnuradio import audio,analogclass my_top_block(gr.top_block):def _init_(self):gr.top_block._init_(self)sample_rate=32000 ampl=0.1 src0=analog.sig_source_f(sample_rate,analog.GR_SIN_WAVE,350,ampl)src1=analog.sig_source_f(sample_rate,analog.GR_SIN_WAVE,440,ampl)dst=audio.sink(sample_rate,)self.connec

22、t(src0,(dst,0)self.connect(src1,(dst,1)if _name_=_main_:try:my_top_block().run()except KeyboardInterrupt:passOpen-source framework for SDR and signal processingBlock-based dataflow architecture“Flowgraph”Each block runs in own threadSignals normalized between-1.0 and 1.0Running C+and Python Under th

23、e hoodGNU Radio Companion is a GUI Tool on top of GNU RadioSource hosted on GitHub https:/ Main Pagehttps:/www.gnuradio.org/GNU Radio Out of Tree Moduleshttp:/www.cgran.org/What is GNU Radio?C/C+Operating SystemMature Windows supportLittle Linux supportImmature Windows supportMature Linux supportLic

24、ensingPaid model,may have site license from test engineering or universityFree and open sourceProgramming SkillsSimpler getting startedGraphical programming new to many SDR developersFamiliarity Several tool chains(C/C+,VHDL,Python)Which Software tool is right for you?User ApplicationCrossbarIngress

25、 Egress InterfaceUSRP Hardware Driver(UHD)Radio CoreHOST PCUSRP FPGAComputation EngineUser DefinedComputation Engine User DefinedUSRP-RF Network On Chip(RFNoC)Industry Update6G TrendsEmbedded,Trustworthy AINetwork OptimizationNext Generation MIMOOpen RAN Evolution6G300 GHz5G956.6G4G24Spectrum Expans

26、ion7 24 GHz“FR3”Sub-THzNew ApplicationsNon-Terrestrial NetworksIntegrated Comms and SensingAn Introduction to 6GRoad From 6G Research To 6G StandardMar 2025:6G workshop202720282019202020212022202320242025202620292030Industry researchIndustry researchRel-16Rel-17Rel-18Rel-19Rel-20Rel-21Rel-223GPP spe

27、cification development5G deployments6G deployments5G-Advanced deployments3GPP specification development3GPP specification development5G-AdvancedRel-21:start with first 6G specifications in 2027Rel-20:kick-off 6G technology study in H2/2025NI Wireless Research Program has enabled critical research si

28、nce 2010Spectrum ExpansionFR3 bandSub-THz bandSub-THz Prototyping and TestingRadio PrototypingReal-time RF data link to FPGA co-processorOpen FPGA for custom signal processing,encode/decode,filtering,etc.Parametric TestingLOVSTFPGAOTAWave Guide InterfaceWR6.5CCD-M12WG BP FilterTX IF 0.1 to 20 GHzRX

29、IF 0.1 to 20 GHzLevelingDUTWave Guide InterfaceWR6.5CCD-M12WG BP FilterTX IF 0.1 to 20 GHzRX IF 0.1 to 20 GHzLevelingLOVSTStimulus/Response characterization of devices and systemsMeasurements:Spectrum,modulation quality,power measurements,waveform research,channel sounding,CW,multi-tone,etc.PA,LNA,F

30、ilter,DSA,OTA Link8881 ControllerWin or RTVST3VST3790379037903(optional)2x2 BB4GHzDigitalPCIe Gen3BackplaneRF UnitRF UnitSingle 3U PXIe Chassis w/18-core Xeon Controller3rd Generation VSTIF frequency coverage up to 54 GHzUp to 4 GHz IBWGeneration and AnalysisPXIe-7903 FPGA CoprocessorVirtex Ultrasca

31、le VU11P FPGA25 Gbps per lane MGT ratesOff the shelf frequency extensionD-band by default(110 GHz to 170 GHz)Other frequency ranges feasibleResearching calibration techniques at sub-THzWaveform research and prototyping via real-time streaming interface to FPGA co-processorStreaming and Measurements

32、in sub-THz frequencies9 CC x 400 MHz,5G NR at 144 GHz,120 kHz SCSOAI 5G End-to-end Reference ArchitectureRich,open-source reference code for core network,gNB,and UE available based on OAI 5G-NR protocol stack Open architecture for real-time 5G end-to-end network to enable research and demonstration

33、of 6G candidate technologies NI-validated system configuration including third-party components to ensure stable performance Detailed documentation,accelerating the set-up time for 5G end-to-end communication systemsOAI Reference Architecture Hardware and SoftwareOAI gNBCore NetworkUHDOAI CN(+PDN Se

34、rver)COTS UEOAI software UEL3L2L1L1L2L3RFRFI/QIPSoftwareall on x86Hardwareall on x86(docker containers)Closed 5G UE StackRFOAI UECOTSUEAppSMA Cables/AntennasAppall on x86UHDUHDI/QOAINIOAIDriverClosed SWAppUHDSMA Cables/AntennasIPModem UEUEIntegrated Sensing and Communications(ISAC)or Joint Communica

35、tions and Sensing(JCAS)Integrated Sensing&Communication ResearchSystem&Protocol designData fusion SynchronizationCross-node schedulingCross-link interferenceCommunications performanceLatencyThroughputSelf-interference cancellerBandwidthFrequencymMIMODynamic environmentsProcessing&delayCircuit overhe

36、adTXRXDSPSISISensing performanceMotionDirectionAccuracyPresenceLatencyTypeChannel modelsSpace,time,frequencyDoppler,delay,RCSResolution IntegrationSystem levelCommon RFCommon waveformKey applicationsTraffic controlMan-machine/XRVital signsNetwork twinning What The signal processing of radar communic

37、ation integration includes integrated waveform design,joint signal transmission,and joint signal reception.The goal is to find a suitable signal waveform to complete the functions of information transmission and target detection at the same time.The NI platform acquires measured data for signal proc

38、essing,model training,and inference.Why choose NI SDR?Open software-defined radio platform with support for custom waveformsFrequency band coverage from sub-6G to THzProvides a wide range of communication protocols(IP,LTE,WLAN,and 5G NR)for easy modification and redefinition.Support continuous colle

39、ction without losing pointsSensing Waveform ResearchAbove:Radar(FMCW)Below:Communications(OFDM)Prototyping System consisting of:JC&S-enabled TransmitterJC&S TXRadar RXComms ReceiverCapabilities:2 GHz real-time bandwidthIntegration with Python-based simulation and prototyping environmentAdaptability

40、to different FR(baseband,IF and RF options)Dr.Maximilian Matth,Barkhausen Institut,Dresden,Germany“Using this setup,we can validate diverse research on mmWave antennas and signal-processing algorithms with real-world signals.Seamlessly,we have the entire Python ecosystem at hand,including link-level

41、 simulators such as HermesPy.”JC&S Prototyping at Barkhausen Institute in Dresden,Germanyhttps:/ Intelligent Surfaces(RIS)In the evolution of 6G,the prospect of using reconfigurable intelligent surfaces(RIS)to assist wireless communication systems has received a lot of attention from academia and in

42、dustry.The main principle is to realize a controllable electromagnetic field through intelligent control.A RIS can control any one or combination of the following:AmplitudePhasePolarizationFrequencyRIS IntroductionXXRIS ApplicationsWireless relay design:By using the unconventional characteristics of

43、 RIS metamaterials,the incident electromagnetic wave can be controlled in any phase in real time,so as to intelligently reconstruct the wireless propagation environment between transceivers and effectively improve the performance of the wireless transmission system.Channel coverage enhancementsReduc

44、e base station power consumptionBeamformingLink optimizationSuccessfully built an RIS wireless communication system that supports QPSK,8PSK(6Mbps),and 16QAM(20Mbps)using the NI SDR platform.RF Signal Processing:NI USRPReal-time control of RIS:NI FPGAs At present,the research team at Southeast Univer

45、sity has upgraded the research to the high-order modulation and mmWave band of 256 QAM.He has also successfully published in the prestigious journal National Science Review.Success Stories-Southeast UniversityResearch results：“A wireless communication scheme based on space-and frequency-division mul

46、tiplexing using digital meta surfaces”https:/ AIApplication Areas of AI/ML in RF WirelessWireless/Mobile CommunicationsNetwork Level:Data flow management,network parameter optimizationMAC:Time/Frequency/Spatial resource scheduling(spectrum sharing),(mmWave)Beam acquisition/selection&trackingPHY:Chan

47、nel estimation&equalization,symbol detection,channel en/decodingRF:Spectrum sensing,Digital pre-distortion Other Applications RF Fingerprinting/Security IP/RRCPDCPRLCMACPHY-HighPHY-LowRF+AntORAN 7.2 SCF FAPI 3GPP F1ORAN CUORAN DUORAN RUUEgNBgNBCore NetworkNo widely re-usable reference data sets avai

48、lableNo common tool sets,very heterogeneous data&metadata formats,missing or incomplete scenario descriptionsOften very manual process to generate data sets with comprehensive metadata(scenario description)Three Challenges of AI/ML in Wireless CommsBridging the gap between theoretical and practical

49、issues around 5G system deployment and implementation for enabling engineers and researchers to rapidly develop and test novel use-casesAI/ML Reference Architecture with NI USRPCustom Application SoftwareFPGAFPGACPUSFP+/SFP28SFP+/SFP28NI RF Data Recording API Application NoteNeural Receiver ExampleU

50、sed publicly available neural receiver model for OFDM-based wireless communication systems(DeepRx)Example to study new validation and testing methodologies on embedded AI in 6G wirelessInitial step:Implemented,trained and validated neural receiver model in link level simulation chainClassical Receiv

51、erChannel EstimationChannel EqualizationSymbol Detection/DemappingCompare&Benchmark ResultsNeural receiver(DeepRx Example)Benchmarking in Simulation ChainSource:Honkala et al.:“DeepRx:Fully Convolutional Deep Learning Receiver”DeepRx Neural Receiver ModelNeural Receiver Real-Time Benchmarking SetupO

52、pen-Air-Interface as system level emulation IP using Open-RAN architecture and running on standard CPU/ServerNeural receiver running on Nvidia GPU and integrated into Open-Air-Interface through Tensorflow RT C-APINI VST&PXI based channel emulator Linux server 2Linux server 1(with NVIDIA GPU)Real-Tim

53、e Channel EmulatorTest Execution EntityBase Station Control&Measurement accessWireless ChannelScenario ControlUser Terminal ControlTest Orchestration Entity-Test scenario configuration and selection-Reception&visualization of test resultsReal-Time User Terminal Emulator(Open Air Interface)NI USRP SD

54、RNI USRP SDRNI PXI systemReal-Time Base Station Emulator(Open Air Interface)Traditional UL PHY ReceiverNeural UL PHY ReceiverRAN controlRAN measurementswireless channel scenario controlLive Demo At NI BoothInitial Neural Receiver Benchmarking ResultsValidation scenario:Single 5G NR link between 1x g

55、NB and 1x UESwitch between gNB uplink traditional 5G NR receiver vs neural receiver Increase signal-to-noise ratio(SNR)for fading channelMonitor uplink block error rate(BLER)Test different Modulation&Coding schemes(MCS)gNB and UE configuration:Uplink/Downlink TDD Bandwidth mode:40 MHz/Subcarrier Spa

56、cing:30 kHzUsed uplink bandwidth:6 PRB=2.16 MHzMCS 11(16QAM,CR 0.37)MCS 15(16QAM,CR 0.6)MCS 20(64QAM,CR 0.55)1.3dB3.2dBSNR dB4 6 8 10 12 14 16 180.010.11UL BLERNeural RX,MCS11Trad.RX,MCS 11Neural RX,MCS 15Trad.RX,MCS 15Neural RX,MCS 20Trad.RX,MCS 20UL BLER vs.SNR(UMi short,TDL-D,speed 2m/s,DS 45ns)I

57、nitial Key Learnings&Takeaways Due to strict timing requirements in cellular wireless communications complexity reduction of AI models is needed to meet timing constraints(in our case 10 x)Low complexity AI models are more difficult to train for a broad variety of wireless scenarios because of reduc

58、ed generalization capabilities Broad testing of embedded AI/ML models is important to detect and iteratively improve for unexpected failure cases using synchronized data Satellite Communication/NTN Prototype TestbedAppAppNetworkNetworkDVBMacSatellite Link5G NR/LTEMacMacMacPHYUSRP/PXIChan EmuUSRP/PXI

59、PHYUSRP/PXIPHYPXI DigitalPHYPXI DigitalPHYUSRP/PXIChannelUSRP/PXIPHYUSRP/PXIOpt CommGatewaySatelliteUENI HWDIY3rd PThe NTN Prototype Testbed on NI SDR PlatformGPS北斗TT&CUSB/DSSSHDRDVB S2/MF-TDMA5G NR/5G NTNChannel EmulatorWide Band RF Record&PA Comprehensive Wireless Research Testbed5G End-to-end SystemmmWave&THz testbedMIMO PlatformGeneral PurposeResearch OAI 5G NR FR1/FR2 ORAN O-RU Emulator mmWave Transceiver sub-THz testing 4x4 MIMO Massive MIMO Transmit,Record andPlayback關注NI微信公眾號【恩艾在您身邊】會后獲取技術演講講義及更多干貨內容