LTE network testbed with USRP and general purpose PC EECRT2 - - PowerPoint PPT Presentation

LTE network testbed with USRP and general purpose PC

EECRT2 Project Kalle Ruttik

Contributions by: G.M. Crespo,J. Kerttula, C. Guo, Y. Beyene, N. Malm

27.11.2013 EECRT2

Department of Communications and Networking Aalto, School of Electrical Engineering

EECRT2 Cognitive Radio test-bed

Purpose Project creates a “living lab” cognitive radio test-bed

– Living lab: Transmission of the real application data over the air interface – Test-bed is a realistic radio network – Test-bed is designed for investigating RRM algorithms

• Test-bed used in

– TEKES EECRT project – One of METIS official test-beds

Test-bed general properties

• Composed of 24 USRP nodes

– USRP N200+SBX: 0.4 – 4.4 GHz – License to transmit in 620 – 650 MHz

• Implementation of TDD-LTE type PHY in software

– BB processing in C++ – RRM in Python

• Currently BB and RRM not combined
• Implementation of BS and UE units

– Two-directional TDD communication

• Currently under the test

– Can address users – Can allocate resource blocks

SDR implementation

Software platform operates as a radio system simulator

MAC

scheduler

BB RRM Data Buffer MAC

scheduler

BB RRM Data Buffer Python C++ Buffer is a wrapper around UHD driver

• It hides synchronization

related issues from the rest of the code

• Handles TDD direction

switching

IP switch

Operates as HW in a loop simulator

MAC

scheduler

BB RRM USRP Data Buffer MAC

scheduler

BB RRM USRP Data Buffer Python C++ Buffer is written as a wrapper around UHD driver

• It hides synchronization

related issues from the rest of the code

• Handles TDD direction

switching

RF AD/DA RF AD/DA

Software RAN in VM

• Implementation of TDD LTE physical layer in software

– Can run Radio Access processing in virtual machine (VM) – Separation of BB processing and sending data to RF – Can be used in servers with remote radio heads

Host OS VM OS BB VM OS BB VM OS BB RF AD/DA

Software architecture

PHY Implementation

• Bit exact DL PHY

– PHY – PSS, SSS – PCFICH – PHICH – PDCCH

• UL PHY

– PUCCH – PUSCH

SDR with multiple USRP units

Using SDR with multiple USRPs

• Using multiple 1 Gbit Ethernet ports for connecting

multiple USRP to a “server”

– The USRP units do not have common clock

• Problem

– Frame synchronization?

• The USRP N200 units do not have global clock
• The packets are transmitted at the arbitrary times

– Clock drift

• Over the time one transmitter will have have one sample more than
• ther

Synchronizing multiple transmitters

• Use software for synchronizing the transmitters

– 2 Tx and 1 Rx connected to same computer – Rx receives both TX signals and computes correction factor – The clocks of transmitters are continuously adjusted by adjusting the samples

• Initial calibration (synchronization)

– Setting the frames to start at the same time – Use different PSS sequences

• Tracking

– Keeping the transmitters synchronized

N200 vs USRP-2932

• 2.5 ppm TCXO frequency

reference

• 0.01 ppm w/GPSDO option
• 2.5 ppb OCXO
• 0.01 ppb w/GPSDO option
• Sample rate and RF frequency both derived from the same oscillator
• Can not simply shift RF frequency
• Sampling difference gives
• Phase error
• Different amount of samples over time

Test Rx Tx2 Tx1

Tx Timing Mismatch Calibration

Add delay

Correlator 1 Correlator 2

Timing delay estimation

UDP/IP UDP/IP

Tracking process

Tx1 source data Tx2 source data Add delay

Tx process

Once the Rx work station estimates the Tx timing mismatch from the correlators‘ outputs, it indicates the Tx which antenna must delay its transmission by N number

• f samples
• n
• ff
• n
• ff

Frame synchronization

• Addressable

Initial frame location Frame starting after synchro

Comparison of freq. drifts between 2 Tx

N200 USRP-2932

5 10 15 20 25 30 35 0.5 1 USRP-2932 2tx timing drift 5 10 15 20 25 30 35

• 1

1 5 10 15 20 25 30 35 0.5 1 5 10 15 20 25 30 35 0.5 1 time (seconds) 2 4 6 8 10 12

• 5

5 N200 2tx timing drift 2 4 6 8 10 12 2 4 2 4 6 8 10 12

• 5

5 2 4 6 8 10 12 1 2 time (seconds)

Histogram of the frequency drifts 10 s. and one LTE symbol

N200 USRP-2932

1 2 3 4 5 6 x 10

• 6

0.2 0.4 0.6 0.8

Phase (radians) pdf

USRP NI2932 1 2 0.2 0.4 0.6 0.8

Samples in 10 s pdf Mean = 1.6111 Std = 0.50163

1 2 3 4 5 6 x 10

• 5

0.1 0.2 0.3 0.4

Phase (radians) pdf

N200 1 2 3 4 5 6 0.1 0.2 0.3 0.4

Samples in 10 s pdf Mean = 3.4681 Std = 1.12

Performance

• Drift between the transmitters clocks

– Drift figure – Histogram of drift

• Error in one LTE OFDM symbol 66e-6 s

– Histogram of symbol error

Radio link performance measurements

Measurements

• Ongoing measurement campaign for indentifying impact
• f using different DL/UL sub-frame configurations in

different transmitters

• Here: SINR and BER measurements in one radio link

– Using different methods for measuring SINR

DL: Downlink SP: Special subframe UL: Uplink

Measurement campaign

• Interference from outside Tx to

inside Rx

– Two TDD radio links

• one outside one inside

– Measure if transmitters

• Sychronised
• Nonsynchronised

– Performance is measured as SINR and BER in radio links

• Performance is measured per

sub-frame

• Currently the measurement

campaign is going on

SINR measurements

• EVM based measurement

– The channel is feed with coded data – The data is received decoded and encoded – EVM is computed from difference of received data and decoded and re-encoded data

• SINR estimation from the spectrum

– Difference between the pilots based signal power estimate and signal plus noise estimate from the resource elements with data

• RSSI
• RSRQ

Estimation of signal strength at different USRP units

Measured BER on USRP-2932 at 630 mHz

Conclusions

• We have TDD LTE type BS that can operate in as server

– The system allows real time operations – The USRP do not have common clock and they are synchronized over the air

• The software system scales for testing TDD based radio

network

– We can measure and control interference in the designed network