FDM based MIMO Spatio-Temporal Channel Sounder

Graduate School of Science and Technology,Tokyo Institute of Technology

Kazuhiro Kuroda, Kei Sakaguchi, Jun-ichi Takada, Kiyomichi Araki

Motivation

The performance of MIMO communication system depends onthe directional as well as temporal behaviour of channel.

The field measurement data of MIMO channel are strongly requiredto develop and evaluate the MIMO communication systems.

An important difference between MIMO and SIMO channel sounding

the MIMO channel sounder needs some kind ofmultiplexing to distinguish between transmittingantennas.

Tokyo Institute of Technology

MIMO Channel Parameter Estimation

H =∑

iγi(t)e

−j2πfcτia(θri )(as(θ

si ))

T

h =∑

iγia(θr

i ) ⊗ as(θsi ) ⊗ af(τi)

Velocity Vector

DOAs

DODs

normal vectorBS

MS

# i

θri

θsi

ms-element array antenna

mr-element array antenna

Frequency response vector is introduced for the wideband measurement.

⊗ : Kronecker product

Tokyo Institute of Technology

A channel matrix at the center frequency of can be expressed asfcmr × ms H

The channel matrix can be reformulated toan dimensional vector .h

mr × ms × mf Hmr · ms · mf

Tokyo Institute of TechnologyApproach to Multiplexing

By using an analogy with multi-user communication scenarios,we considered three types of multiplexing techniques for soundingpurpose in terms of realtime measurement, hardware cost effectiveness,and major drawbacks.

The three types of multiplexing :

CDM ( Code Division Multiplexing )

TDM ( Time Division Multiplexing )

FDM ( Frequency Division Multiplexing )

It is realizable only by changing onetransmitter to the other antennas witha switch.

・Hardware cost

・Realtime measurement

Measurement which has ms timesbaseband signal period and furthermoreguard interval and switching are needed.

・Major drawback Absolute time synchronizationbetween transmitter and receiveris required.

excellent

poor

…

TDM based TechniqueTokyo Institute of Technology

Transmitter

Transmitter

・Hardware cost

・Major drawback

poor

・Realtime measurement excellent

Measurement period doesn’t dependon ms.

It needs ms transmitter channels.

Dynamic range of the system is limitedby ms due to cross-correlation betweendifferent codes.

Tokyo Institute of TechnologyCDM based Technique

Transmitter

f1 f2 f3・Hardware cost

・Major drawback

good

・Realtime measurement good

Measurement period is ms timesbaseband signal period.

It requires ms local oscillators, but one signal generator.

Some modification is needed forthe data model, since the frequencysample points in each transmittingantenna are different.

Tokyo Institute of TechnologyFDM based Technique

FDM

CDM

TDM

Realtime Measurement

Hardware Cost Major Drawback

good good

excellent

excellent

poor

poor

Comparison of Multiplexing

Frequency shift in Tx signals.

Synchronization between Tx and Rx

Cross-correlation between codes

FDM achieve both realtime measurement and hardware cost effectiveness.

Therefore, we chose the FDM based technique.

Tokyo Institute of Technology

f

f

Tx Antenna 1

Tx Antenna 2

Rx Antenna

DFT

f

f

f

Rate：

Multi-tone FDM

∆f =∆F

ms

∆f

Multi-tone Signal

Tokyo Institute of Technology

∆f

∆F

By using this, the parameter sets can be simultaneouslyestimated.

Channel Impulse Response

aFDM(τi)�= [1, e−2π∆fτi, · · · , e−j2π(ms−1)∆fτi]T

h′ =∑

iγi(t)ar(θ

ri ) ⊗ a′s(ψs

i ) ⊗ af(τi)

a′(ψsi ) = a(θs

i ) � aFDM(τi)

A FDM response vector is defined as

By using this vector, the transmitting array response vector is rewritten as

Therefore, the channel response vector for FDM based MIMO systemis written as

{θri , θ

si , τi}

Tokyo Institute of Technology

� : Hadamard product

Hardware Implementation

5845MHz

A/D DFT

Dow

nC

onve

rt

ES

PR

IT

PC

f

Arbitrary Waveform Generator

880MHz 880.125MHz4970MHz

125kHz

Low IF : 5MHz

5850MHz

5850MHz

Transmitter

Receiver

Tokyo Institute of Technology

S/P

Measurement Environment

Tx antenna

∆F

∆f

500kHz

125kHz

9.5MHz

30 times

about 30dB3-D Unitary ESPRIT

Rx antenna

Bandwidth

Snapshot

SNR

Estimation algorithm

2-element ULA

2-element ULA4m

Tx Rx

Tokyo Institute of Technology

anechoic chamber

Measurement Results

- Calibration error

- Setup error

The factor of the main error:

Setting Value : the every 15[deg] grid

Tokyo Institute of Technology

Conclusion

We implemented the FDM based MIMO channel Sounder.

We confirmed the validity of the FDM based architecture throughmeasurements in anechoic chamber.

Tokyo Institute of Technology

After considerable discussions about multiplexing techniques todistinguish between the transmitting antennas, the FDM based architecture was chosen to achieve cost effectiveness and realtimemeasurement.

In the frame work of FDM, we proposed a new transmitting signalconfiguration and a new algorithm to estimate the MIMO channelparameters.