1.multiuser channel estimation
DESCRIPTION
Block I. Block II. Block III. Antenna. Matrix Products. Data. Multiuser Detection. Decoder. Block IV. Correlation Matrices (Per Bit). Inverse. Detected Bits. A 0 H A 1 O(K 2 N). Delay. M U X. d. Decision Feedback. Multistage Detection (Per Window). Multiple Users. R br - PowerPoint PPT PresentationTRANSCRIPT
IMPLEMENTING CHANNEL ESTIMATION ALGORITHMS ON HARDWARE
Sridhar Rajagopal, Suman Das and Joseph R. Cavallaro
1.Multiuser Channel Estimation1.Multiuser Channel Estimation 5.Task Decomposition5.Task Decomposition
9 10 11 12 13 14 150
0.5
1
1.5
2
2.5
3
x 105
Number of Users
Dat
a R
ates
Data Rates for Different Levels of Pipelining and Parallelism
(Parallel A) (Parallel+Pipe B)
(Parallel A) (Pipe B)
(Parallel A) B
A B
Sequential A + B
Data Rate Requirement = 128 Kbps
3 .Base-station Receiver3 .Base-station Receiver
The Wireless ChannelThe Wireless Channel
Direct PathReflected Paths
Noise + MAI
User 1
User 2
Base Station
2.Third Generation Communication Systems2.Third Generation Communication Systems
6. Exploiting Pipelining and Parallelism6. Exploiting Pipelining and Parallelism
7. Meeting Real-Time Requirements7. Meeting Real-Time Requirements
Multiple Users
Channel
Estimation
Multiuser
Detection
DecoderData
Pilot
Demod-ulator
Antenna
Decision Feedback
MUX
Detected Bits
+
Delay
MUX
d
b (known)
8.Real-Time Implementation8.Real-Time Implementation
VLSI DSP FPGA High Performance Processors
DSP + FPGA
Joint Work with
Praful Kaul (UIUC)
Parthasarathy Ranganathan
Dr. Sarita Adve (UIUC)
SpreadingFactor
Number ofBits / Frame
Data RateRequirement
4 10240 1024 Kbps32 1280 128 Kbps
256 160 16 Kbps
4. DSP Implementation4. DSP Implementation
K
1
Block IV Matrix Products
A0HA1
AHrA0HA0
O(DK2Me)
1
K
Block III
A1HA1
data
d
• TI TMSC6701, projected at 250 MHz.
• 1953 cycles available for detection of 1 bit
assuming data rate of 128 Kbps.
• In-depth profiling to find bottlenecks.
• Multiuser Detection
– needs to performed continuously to meet data
rate requirements
• Channel Estimation
– can be updated less frequently
• Single DSP does not meet real-time requirements
• Multiuser Detection - Bottleneck!
• Multiuser Channel Estimation
– Need to know the Channel for proper detection
– Delays and Amplitudes of each user and each path
• Send sequence of known bits (Pilot / Preamble)
– Pilot Code-Multiplexed with Data
– Pilot Time-Multiplexed with Data
• Multiuser Detection
– Use knowledge of channel for reliable detection
•Multiple Users•Multiple Access Interference
•Multipath Delays
•Fading
•Additive White Gaussian Noise
Channel Effects
• Multiuser Channel Estimation Methods
– Subspace
– Maximum Likelihood
• Joint Estimation and Detection
– Computationally Efficient
– Better BER Performance
• W-CDMA - Wideband CDMA (5 MHz)
• 3G Communication Systems
–Integrating Multimedia Capabilities
–Quality of Service (QoS)
–Multi-rate Services
–Higher Data Rates
• 2048,384,144 Kbps
• N - Spreading Code Length
• K - Number of Users
• A - [A0 A1] - Channel Estimate
• D - Multiuser Detection Window
• r - Received bits of K users
– Can be Data or Pilot (with
interference/fading)
• b - Known Pilot bits at the receiver
• d - Detected Data bits
• Data’ - Data synchronized with d
Parameters
Accelerating the blocks in Multistage Detection
to meet real-time requirements.
Graph shows the data rates achieved by different levels
of acceleration for multiuser detection.
Block I Block II Block III
Inverse
Correlation Matrices (Per Bit)
Rbr
O(KN)RbbAH = Rbr
O(K2N)
Multistage
Detection
(Per Window)b
pilot
data
M
U
X
d
data’ M
U
X
RbbAH = Rbr
O(K2N)
O(DK2Me) d
Rbr
O(KN)
Rbb
O(K2)
Matrix Products
A0HA1
O(K2N)
AHr
O(KND)
A1HA1
O(K2N)
A0HA0
O(K2N)
Block IV
Multiuser Detection
Channel Estimation
Time
Task A Task B
Task ATask B