MULTIANTENNA CELLULAR COMMUNICATIONSebjornson/poster_swe-ctw_oct2011.pdf · 2011-10-24 ·...
Transcript of MULTIANTENNA CELLULAR COMMUNICATIONSebjornson/poster_swe-ctw_oct2011.pdf · 2011-10-24 ·...
MULTIANTENNA CELLULAR COMMUNICATIONS
Doctoral Thesis in Telecommunications
Emil BjörnsonKTH, ACCESS Linnaeus Center, Signal Processing Lab
CHANNEL ESTIMATION, FEEDBACK, AND RESOURCE ALLOCATION
IntroductionDownlink Cellular Communications:
- Multiple Transmitting Base Stations - Multiple Receiving Users - Sharing a Common Frequency Band - Limited by Power Constraints - Limited by Co-User Interference
Space-Division Multiple Access (SDMA):
- Multiantenna Transmission - Beamforming: Spatially Directed Signals - Combining: Spatially Directed Reception - Serve Multiple Users - Simultaneously - Controlled Co-User Interference
System OperationSDMA Requires Accurate Channel Knowledge:
- Find Strong Channels, Compatible Users, and Good Beamforming - Achieved By Channel Estimation and Feedback: (colors represents different parts)
Channel EstimationTraining-Based Channel Estimation:
- Send Known Signals in All Spatial Directions - Sufficient to Estimate Channel Behavior - Limited by Interference and Noise
UserBase Station
Training Signaling inall Spatial Directions
Problem: Training Signal Optimization
- Select Training Signal to Minimize Error (Mean squared error of estimate) - Adapt to Channel and Interference Statistics (eigendirections and eigenvalues) Strength of Channel
EigendirectionsStrength of Interference
Eigendirections︸ ︷︷ ︸
EigendirectionsCombined in
Opposite Order
Reference: E. Björnson, B. Ottersten, “A Framework for Training-Based Estimation in Arbitrarily Correlated Rician MIMO Channels with Rician Disturbance,” IEEE Transactions on Signal Processing, Mar. 2010.
Proposal: Explicit Training Strategy
- Based on Statistical Eigendirections - Concentrate on Strong Channel Directions - Strong Directions when Interference is Weak - Adapt to Estimation of Different Quantities
How to Measure System Performance?Many Definitions of Single-User Performance:
- Achievable Data Rate (perfect and long coding) - Bit Error Rate (complicated expressions) - Mean Squared Error (difficult to interpret)
Benchmark: Optimal Resource AllocationEvaluate Practical Strategies towards Optimum
- Most Resource Allocation Problems are NP-hard (e.g., sum performance and proportional fairness) - Computationally Expensive to Solve Optimally - Can be Seen as Search in Performance Region
Practical Resource AllocationProblem: Practical Resource Allocation Strategy
- Select Users and Beamforming - Limited Computational Complexity - Only Require Local Channel Information
Multi-User Performance has Fairness Dimension:
- Divide Available Power among Users - Create and Control Co-User Interference - Illustrated by the Achievable Performance Region:
PerformanceUser 1
Performance, User 2
PerformanceRegion
Upper Boundary
︸ ︷︷ ︸They All Improve with the
Signal-to-interference-and-noise ratio (SINR)Optimizethe SINR
Any Multi-User Resource Allocation Problem:
- Optimal Solution on the Upper Boundary - Difficult to Know Which Point (region is unknown)
Proposal: Branch-Reduce-and-Bound Algorithm
1) Cover Performance Region with a Box 2) Divide the Box into Two Sub-Boxes 3) Find Lower/Upper Bounds in Boxes 4) Calculate Lower/Upper Bounds on Optimum 5) Continue with one Sub-Box at a Time End: When Bounds on Optimum are Tight Enough
Reference: E. Björnson, G. Zheng, M. Bengtsson, B. Ottersten, “Robust Monotonic Optimization Framework for Multicell MISO Systems,” IEEE Transactions on Signal Processing, Submitted in Mar. 2011.
1) Cover Region 2) Branch (Divide)
3-4) Reduce & Bound 5) Continue With Sub-Box
Use forBounds
References: E. Björnson, R. Zakhour, D. Gesbert, B. Ottersten, “Cooperative Multicell Precoding: Rate Region Characterization and Distributed Strategies with Instantaneous and Statistical CSI,” IEEE Transactions on Signal Processing, Aug. 2010.
E. Björnson, N. Jaldén, M. Bengtsson, B. Ottersten, “Optimality Properties, Distributed Strategies, and Measurement-Based Evaluation of Coordinated Multicell OFDMA Transmission,” IEEE Transactions on Signal Processing, To appear.
Proposal: Distributed User Selection
- Select Spatially Separated Users - Coordinate Selection between Cells - Strategy: Greedy Algorithm Exploiting Previous Selections in Adjacent Cells
Proposal: Low-Complexity Beamforming
- Explicit Parameterization of Optimal Beamforming - Hard to Find Optimal Parameters - Easy to Find Parameters Giving Good Performance - Strategy: Beamforming with Heuristic Parameters
Proposal: Dynamic Cooperation Clusters:
- Adjacent Base Stations (BSs) Cooperate - Outer Circle: Coordinate Interference - Inner Circle: Serve with Data - Some Users Served by Multiple BSs
Users Selected by Adjacent Cells
User Selection Within the Cell
Evaluation on Realistic Channel Measurements
- Measured Around KTH Campus (thanks to P. Zetterberg and N. Jaldén) - Optimal vs. Practical Resource Allocation - Confirms that Proposed Strategies Work Well - Important to Coordinate Interference between BSs - Joint Transmission Requires Accurate Synchronization
0 200 400 600 800 1000−700
−600
−500
−400
−300
−200
−100
0
Distance [m]
Dis
tanc
e [m
]
Direction of BSs
Measured Routes
BS 1 (Vanadislunden) BS 2 (KTH Nymble)
Beamforming Controlledby a Few Parameters
Box
Restart When Channel Estimates are Outdated
Training Signaling:Channel Estimation
at Each User
Resource Allocation:Base Stations Perform
User Selection & Beamforming
Data Transmission
Feedback:Send Channel Estimates
to Base Stations
User 1
User 2
Base Station
MultipleData Streams
Quantization and Limited FeedbackFeedback of Channel Information using Limited Number of Bits
- Channel Direction: User Selection, Beamforming - Channel Quality: User Selection, Power Allocation
How to Divide Bits between Direction and Quality?
- Depends on Channel Conditions and Number of Users - Answer: ~3 Bits for Quality and Remaining Bits for Direction - Accurate Direction Information Essential to Control Interference
DirectionQuality
Channel
One or Multiple Streams per Multi-Antenna User?
- Determines Number of Directions to Feed Back - Answer: Only One Data Stream per User - Enables Receive Combining to Reject Interference - Achieves Better Effective Channels
References: E. Björnson, M. Bengtsson, B. Ottersten, “Receive Combining vs. Multistream Multiplexing in Multiuser MIMO Systems,” Proc. Swe-CTW, October 2011.E. Björnson, K. Ntontin, B. Ottersten, “Channel Quantization Design in Multiuser MIMO Systems: Asymptotic versus Practical Conclusions,” Proc. IEEE ICASSP, May 2011.
Cake of Feedback Bits:
Direction: Many bitsTo Control Co-User
Interference
Quality: 3 bits
MultipleData Streams