1. Space-Time Fronthaul Compression of Complex Baseband Uplink LTE Signals
Jinseok Choi, Brian L. Evans and *Alan Gatherer
Embedded Signal Processing Laboratory
Wireless Networking & Communications Group
The University of Texas at Austin
*Huawei Technologies, Plano, Texas

2. Fronthaul Link and Challenge
RRH: Remote radio head
BBU: Baseband processing unit UE: User equipment
Network Features
Separation of RRH and BBU
Physical link between RRH and BBU
Multiple RRH support at one BBU
Challenge
Rapidly growing data traffic
Very expensive fronthaul links
RRH BS
BBU UE
ex. Cloud Radio Access Network
Fronthaul Links
DL + UL Monthly Data Traffic
[Ericsson, Akamai, 2013] 2

3. Key Intuition Our Contributions
Dimensionality reduction for massive multiple receive antennas
Adaptive quantization of dimension-reduced signals
Our Contributions
Space-time fronthaul compression method for uplink LTE signals
High compression ratio
Noise reduction: communication performance improvement
Numerically validate the proposed compression method

4. System Model Network Model Signal Model RRH Process
Single-Antenna UEs
M Antennas/RRH
RRH
Process
RRHs: M Rx antennas
User: single antenna
Channel: frequency selective
No inter-cell interference
RRHs - equipped with massive antennas
RRHs - serve multiple users
Users - equipped with a single antenna
In our network model,
In our signal model,

5. Compression Process Proposed compression method Compression at the RRH
Time domain compression for complex baseband LTE uplink signals
Decompression at the BBU in a reverse order
To resolve the difficulties, key math tools
PPP: explains the distribution of these points.
Assuming UE at the origin, pdf of the closest base station: Distance r
r is a random variable

6. Principal Component Analysis
M: # antennas at RRH
N: # samples per stream
L: rank of the matrix Y
Principal Component Analysis (PCA)
Low-rank approximation
Dimension reduction (DR) by
SNR gain due to denoising
Compression Block
Bit-Allocation Q2 QL
PCA
Rank Search Q1
Linear Transform
SVD
Compression Rate with DR
Original # of samples
Reduced # of samples 6

7. Transform Coding Transform Coding with Bit Allocation (BA)
M: # antennas at RRH
N: # samples per stream
L: rank of the matrix Y
bSD: standard quantization bits
bi: quantization bits for ith p, v
Transform Coding with Bit Allocation (BA)
Performs individual quantization pi and vi
Minimizes overall weighted mean-squared quantization error
Uses a simple greedy algorithm
Mean-squared error of Q(pi)
Compression Block
Bit-Allocation Q2 QL
Transform Coding
Rank Search Q1
Linear Transform
SVD
i-th eigenvalue
Compression Rate with BA
Original # of total quantization bits
Compressed # of total quantization bits 7

8. Compression Rates Compression Rate with DR Compression Rate with BA
Compression Block
Bit-Allocation Q2 QL
PCA
Transform Coding
Rank Search Q1
Linear Transform
SVD
Compression Rate with DR
Compression Rate with BA
Compression Rate with DR+BA
M: # antennas at RRH
N: # samples per stream
L: rank of the matrix Y
bSD: standard quantization bits
bi: quantization bits for ith p, v
Specifying the multi-user jt beamforming, we employ the distributed zero-forcing. The signal model for this is is shown here, and we can see that the inter-user interference is nullified by zero-forcing beamforming, and the desired channel gain is modified by the beamformer. We assume the sum-power constraint.
Quantization side information 8

9. Validation – Link Level Simulation
Simulation Setting
Parameters for LTE Transmission
64-QAM modulation
64 antennas
{4, 8} users
Resource blocks per user
: {12, 6} blocks each
Compression block length
N = 1096 ; (1024+CP)
Pedestrian A channel model
: Four delay paths
Transmission BW [MHz]
1.4 3 5 10 15 20
Occupied BW [MHz]
1.08
2.7
4.5
9.0
13.5
18.0
Guardband [MHz]
0.32
0.3
0.5
1.0
1.5
2.0
Sampling Frequency [MHz]
1.92
3.84
7.68
15.36
23.04
30.72
FFT size
128
256
512
1024
1536
2048
# of occupied subcarriers 72
180
300
600
900
1200
# of resource blocks 6 25 50 75
100
# of CP samples (normal)
9 x 6
10 x 1
18 x 6
20 x 1
36 x 6
40 x 1
72 x 6
80 x 1
108 x 6
120 x 1
144 x 6
160 x 1
# of CP samples (extended) 32 64
384
We performed link-level simulation for the validation of our algorithm for 10MHz case
the simulation is set as 64QAM Modulation with 8, 16, 32 and 64 antenna cases. And the number of users is fixed as 4.
Assigned Resource blocks per user is 12 block, total 48 block out of 50 blocks (which is 96% of full usage).
Compression Block Length is 1096 and Pedestrian A channel model is used.
[Fundamentals of LTE, Arunabha Ghosh, Jun Zhang, Jeffery G. Andrews, Rias Muhamed, 2010] 9

10. Validation: Error Vector Magnitude
3 dB Gain
6 dB Gain
Achieves 8.0x / 5.0x compression for 64-antenna cases with 4 / 8 users
Achieves 6 dB / 3 dB SNR gain – EVM improvement
Satisfies error vector magnitude (EVM) requirement for 64-QAM (< 8%)
Numerical Results 10

11. Validation: Bit Error Rate
6 dB Gain
3 dB Gain
Achieves 8.0x / 5.0x compression for 64-antenna cases with 4 / 8 users
Achieves 6 dB / 3 dB SNR gain – BER improvement
Numerical Results 11

12. Conclusion Space-time fronthaul compression method Limitations
Achieves 8.0x / 5.0x compression for 64-antenna cases with 4 / 8 users
Achieves 6 dB / 3 dB SNR gain – BER improvement
Satisfies EVM Requirement for 64-QAM (< 8%)
Limitations
Compression ratio depends on # antennas, # users & channel state dimension
Low rank matrix is necessary for high compression ratio
Future work
Analyze error performance
Develop downlink space-time compression method