multicell uplink spectral efficiency of coded ds- cdma with random signatures by: benjamin m....

MULTICELL UPLINK SPECTRAL EFFICIENCY OF CODED DS- CDMA WITH

RANDOM SIGNATURES

By: Benjamin M. Zaidel, Shlomo Shamai, Sergio Verdu

Presented By: Ukash NakarmiUniversity of Houston

Academic Advisor: Dr. Zhu Han

Outline:• Introduction• System Model• Spectral Efficiency• Power Allocation Policies• Conclusion

INTRODUCTION

• Multi Cell Uplink communication Model is Suggested

• Comparative study of Spectral efficiency for Different Multi User detection

• Compares Four Detection Techniques :• Matched filter detection • Single Cell Optimum Detector• MMSE• MMSE- Successive Interference Cancellation( MMSE-SC)

• Under Three Power Allocation Models:• Equal Power Policy• Equal rate Policy• Maximum Spectral Efficiency Policy

• CDMA System with Random Spreading Sequences is Examined

• Scenario:• Linear Cell Array Model• Number of users and Processing gain Goes to Infinity• System Load: finite Constant

Where Do we Use Random Matrix Theory

• In the Performance Measurement • Signal to Interference Plus noise ratio• Spectral Efficiency• Converge to Deterministic Value

• Performance Measurement are function of Eigen values Distribution Of Random Matrices

• Converges with our Assumption of Numbers of User going to Infinity.

• Uses Stieltjes Transform

?

SYSTEM MODEL

• Assumes linear Cell Array Model• Given by Wyner’s Cell Model

Fig: Linear Cell Array Model

• Yi=Signal vector received at Arbitrary cell at Discrete time related to ith symbol

• Xi= [x1i, ……………, xKi]

•Denotes Vectors of Symbol from the Users Operating in Adjacent Cells.•The symbols are iid Gaussian : Capacity Achieving parameters:

• S are the NxK Matrix• Columns are N chip long Random spreading

signature• K is the Number of Users in the Cell

considered• Power Allocation:

• Same power Allocation Policy is Applied to All Cell.• Power Assignment Function:

Constraints:

SPECTRAL EFFICIENCY

• Performance Measure parameter: Spectral Efficiency

gk is Total number of bits per chip for k user.As: K-> infinity, g(k/K)-> g(x)

Matched Filter

• Passes Received Signal• Treats all Interfering Signal as AWGN• Converges to our Assumption K,N-> Infinity,

B= Constant• Multiuser Efficiency:

Where E{ H(p)} is the Expectation of Limiting function H(P) to which Distribution Of Received Power Converges.

Spectral Efficiency:

Minimum Mean Square Error Detector

• Passes Received Signal• Minimizes Mean Square Error• With same K, N Assumption:

Where, ȵms = Multiuser Efficiency for MMSE

SINGLE CELL OPTIMUM DETECTOR

• Uses relation Between Optimum Multiuser detector and Linear MMSE

MMSE Successive Interference Cancellation:

• Has Linear MMSE at Each Stage• From First User in Cell keeps on Cancelling

Interferences• So Multi User Efficiency for Each User within

Cell also Differs• Spectral Efficiency:

POWER ALLOCATION POLICIES

• EQUAL POWER :

For Matched Filter:

FOR SCO:

Fig: Spectral Efficiency with Equal Power• MMSE-SC is has Optimal Efficiency• As E/N increases MMSE surpasses SCO• But if Codes of Adjacent Cells are known ,

Then MMSE- SC is no more Optimum.

Fig: Comparison Of Equal Power and Equal rate Allocation

• For Low E/N Ratio Equal Power and Equal rates have comparable Spectral Efficiency

Fig: Spectral Efficiency: for Optimum

For Low Eb/No Matched Filter and SCO has More Spectral Efficiency than MMSE AND MMSE- SC

CONCLUSION

• Analyzed Spectral Efficiency of Four Multi User Detector

• Comparison Of Power Allocation policy• Considers Simple Linear Array Cell • Can be implemented for two dimensional

Hexagonal, Multi Cell Model

• QUESTIONS ??

THANK YOU