### Mid Semester Presentation - High Speed Digital Systems Lab

```Technion
Israel Institute of Technology
Spring 2013
Mid Presentation
Supervisors:
Rolf Hilgendorf, Debby Cohen
Consultant:
Eli Shoshan
Students:
Etgar Israeli, Shahar Tsiper

Theory

Project Definition and Goals

Project Main Stages:

◦
Matlab Reconstruction
◦
AWR Activities – Part A
◦
AWR Activities – Part B
◦
A-Matrix Calibration
◦
MWC Development Support Systems
Epilogue
~
~

Multiband model:
• N – max number of transmissions
• B – max bandwidth of each transmission

Goal: Blind detection + Recovery

Minimal achievable rate: 2NB << fNYQ
nTs
1
2Ts
m
sequences Expander
q
m
sequences
pm/ q (t )
nTs
fp
1
2Ts
ym[n]
Δ
q
fs
fp
• Support S recovery
• Signal reconstruction
zS  f
~
y f 
~
 AS

z f 
zS  f   AS† y  f 
~
m sequences
Reconstructor
~
Support
Recovery
ym[n]
~
zS  f   AS† y  f 
AS
~
A
z f 
zS  f 
z f 



In theory there is a solid algorithm for building the
A-Matrix. We use the fourier coefficients of the
mixing series:
We’re interested in finding the coeff. cil Therefore
we’ll use:
We can further simplify if the mixing series are step
functions:


We can now define an all constant A-Matrix:
We can now use the same A matrix in time
domain. Due to the invariance for iDTFT.

After the Support Recovery process:
A  AS
m L , mr


r 2 N
Using Moore-Penrose psuedo-inverse
process for the matrix:
Solving the problem:

Matlab reconstruction algorithm

AWR Activities

A-matrix Calibration

MWC development support systems
(Labview programming Rolf/Idan)

Understanding and fixing the Matlab code

Learning AWR tool and Modeling MWC

Deeper understanding of the main issues the system
suffers from

Developing calibration solutions for the system

Implementing the solutions on the actual system
◦
Matlab Reconstruction
◦
AWR Activities – Part A
◦
AWR Activities – Part B
◦
A-Matrix Calibration
◦
MWC Development Support Systems

We’ve developed signal comparison algorithm
using cross-correlation.

Main Issues:
◦ Support recovery is successful at approx. 80% of the
runs (better % for qpsk than sinc)
◦ If  ⊂  the recovery adds redundant harmonics
◦ If  =  time reconstruction still isn’t perfect

Understand schematics of analog part of
new MWC

Get understanding of AWR tool

Define method for input and output files
◦ Matlab , CSV etc.

Enter first draft of MWC schematic


Refine MWC design
◦
Get final spice models for all components
◦
Get model of card
◦
Enter final schematic
◦
Ensure synchronization between patterns
◦
Ensure synchronization with trigger
◦
How to create the input scenarios (AWR or matlab)
◦
Sampling rate for AWR simulation and for output
Basic Verification of output data using matlab
◦
Is input mapped to output as expected
◦
Limits for input signal (saturation, undetectable due to noise)
◦
Anti-aliasing filter response

Understanding the Physical Issues

Using the AWR model output define AMatrix
◦ Perform developed procedure using model and
matlab only
◦ Perform procedure using MWC development
systems described below

Phase Shifts inside the system:
◦ Signals enter with unknown phase into the analog card. We should make
sure we know how to recover the signals with their original phases.
◦ Analog Low-Pass Filter causes unknown phase shifts between the different
channels.
◦ Fixed phase shift between the mixer channels and the Expander Unit.

Noise Sources:
◦ Impedance mismatches in the input cable end – attenuator is used, and acts
as a noise source.
◦ Analog splitter before entering the different mixers provide as a noise
source.
◦ Analog Low-Pass Filter causes noise.

Modeling each part of the system
independently, according to schematic

Trying to develop specific solutions to each
of the micro-problems
Unknown
?
Attenuator
Noise
nTs
1
2Ts
m
sequences Expander
q
ATT
m
sequences
pm/ q (t )
nTs
fp
1
2Ts
ym[n]
Δ
Unknown
phase
Splitter
Noise
LPF – Noise &
phase shift
Phase
shift

Multiplying by a correction matrix before applying the
original A-Matrix - 1 .
◦ In order to get 1 we planned to drive an impulse function into the
system, and determine the impulse response for each Hardware
Channel

Applying a filter after multiplying the signal with the AMatrix - 2
◦ We’ll use multiple known fixed carriers inputs (modulated sincs or
simple sine waves) in order to devise the required 2

Thinking on new calibration methods after
examining a full analog model or real MWC
System - Still work in progress

Synchronizing the A matrix’s via cyclic shifts
to the mixer series - Might be necessary

Data acquisition using NI converter with
external sampling clock

Immediate system based on Tabor AWG
◦ Load data from AWR simulation

Final development system using NI AWG
◦ NI sync card and external clocking

Matlab:
◦ Used for full modeling of the MWC system –
Already given – need to be fixed
◦ Calibration Methods

AWR:
◦ Implementing an analog model of the entire
MWC system.
◦ Linking the analog AWR frontend and the
digital Matlab backend

Labview:
◦ Implementing calibration procedure
Main missions
Fix Matlab reconstruction
algorithm
Understanding the existing Matlab
Becoming proficient in AWR
environment
Understand schematics of analog
part of new MWC
Define method for the input and
output betweem AWR amd Matlab
Enter first draft of MWC schematic
Entering second stage of project:
Refine MWC design
week1 2/6
week2 9/6 week3 16/6 week4 23/6 week5 30/6 week6 7/7 week7 14/7
Technion
Israel Institute of
Technology
Spring 2013
Mid Presentation
Supervisors: Rolf Hilgendorf, Debby Cohen
Students: Etgar Israeli, Shahar Tsiper
```