Brigham Young University Homepage

Digital Communications: A Discrete-Time Approach
by Michael Rice

A Complete QPSK System

Introduction

The previous QPSK Simulink exercises assumed perfect synchronization, perfect timing synchronization, or perfect phase synchronization. It is now time to explore how it is in practice--neither carrier phase nor timing is known to the detector and must be estimated. Simultaneous carrier phase and symbol timing synchronization is an example of a joint estimator: more than one unknown parameter is estimated. This is accomplished using two PLLs, one for timing synchronization (the inner loop) and one for carrier phase synchronization (the outer loop). The issue of carrier phase ambiguity is still with us, so differential encoding is also used here.

In this exercise, you will design a joint carrier phase ans symbol timing timing synchronizer for QPSK based on two PLLs. There are three sample rates in play in this design, so you must be careful in assigning sample times to the Simulink blocks. The detector, with its accompanying bit timing synchronizer, will be used to process the modulated sampled in the file qpskcompletedata.mat.

Textbook References

M-ary QAM: Section 5.3 (pp. 238 - 260), discrete-time realizations: Section 5.3.2 (pp. 256 - 260), partial response pulse shapes: Section A.2 (pp. 682 - 687), discrete-time PLL: Section C.2 (pp. 732 - 747), carrier phase synchronization for QPSK: Section 7.2 (pp. 365 - 375), differential encoding for phase ambiguity resolution for QPSK: Section 7.7.1 (pp. 402 - 405) general discussion of timing synchronization: Section 8.1 (pp. 436 - 438), discrete-time techniques for symbol timing synchronization for binary PAM: Section 8.4 (pp. 445 - 494), discrete-time techniques for symbol timing synchronization for MQAM: Section 8.5 (pp. 494 - 497).

Specifications

normalized sample rate: 8 samples/bit
normalized carrier frequency: 0.3 cycles/sample
carrier phase unknown!!
average energy: 1
pulse shape: SRRC (50% excess bandwidth, span = 12 symbols)
symbol clock offset: unknown!!
input file qpskcompletedata.mat
phase ambiguity resolution differential encoding

bits   phase shift
------------------
 00     0 deg
 01    90 deg
 10   -90 deg
 11   180 deg
								
packet format: PRE = 2000 bits (you don't need to know what they are)
SYNC = 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1
DATA = 3206 bits

Preliminary Design

Design the Detector

Design the detector, shown below, using blocks from the Simulink, DSP System, and Communications System Toolboxes.



Design both loop filters to create a second order loops to the following specifications: closed-loop equivalent noise bandwidth = 0.01 (normalized to the symbol rate), damping factor = 0.7071.

Design Notes

  1. Because I plan on using the zero crossing detector for the timing error detector (TED) and the zero crossing TED operates at 2 samples/symbol, the matched filter output output is downsampled by 4. The matched filter and its input operate at 8 samples/symbol: the downsampled sequence is at 2 samples/symbol.

  2. The carrier phase synchronization loop is the outer loop shown in red.

    • In the two previous two Simulink exercises devoted to carrier phase synchronization for QPSK (Carrier Phase Synchronization for QPSK Using the Unique Word and Carrier Phase Synchronization for QPSK Using Differential Encoding/Decoding), perfect timing synchronization was assumed and this assumption allowed the carrier phase PLLs to operate at one sample per symbol.

    • In contrast, the matched filter outputs here are at 2 samples/symbol. Consequently, the CCW rotation block operates at 2 samples/symbol.

    • The PED operates at 2 samples/symbols, but only updates at 1 sample/symbol. This is why the PED needs to strobe input. So, when the strobe is asserted, the PED output follows the formula (7.26) on page 370. When the strobe is not asserted, the PED output is 0. (For those of you familiar with multirate discrete-time processing, this is like an upsample-by-two operation.)

    • The loop filter and DDS operate at 2 samples/symbol. (In the context of multirate processing, the PED does the upsample-by-two and the loop filter is the "interpolating filter" to smooth out the inserted zeros.)

  3. The symbol timing PLL is the inner loop shown in blue. Except for the CCW rotation block, the symbol timing PLL is identical to the one you designed for the Symbol Timing Synchronization for QPSK Simulink Exercise.

    • Use a Farrow interpolator. You can use either the piecewise-parabolic version (see the top of Figure 8.4.16 on p. 471) or the cubic version (see the bottom of Figure 8.4.16 on p. 471).

    • The strobe signal is derived from the decrementing modulo-1 counter. Use the underflow condition in the decrementing mod-1 counter as the strobe signal.

    • You need to create an "enabled" version of the zero-crossing TED. When the strobe is high, the TED output is given by equation (8.37) on page 456. When the strobe is low, the TED output is zero.

    • For interpolation control, use the decrementing modulo-1 counter described in Section 8.4.3 (pp. 475 - 477).

    • The "enabled hold" block outputs the previous value when the strobe is low, or passes the input to the output when the strobe is high. This block is needed because the fractional interval is updated by the mod-1 counter only once per symbol, but the interpolator operates at 2 samples/symbol. The hold operation ensures that the interpolator is using the proper value for the fractional interval.

      Simulink provides a skeletal enabled subsystem:

      Simulink -> Ports & Subsystems -> Enabled Sybsystem

      The default enabled subsystem is shown below



      The subsystem consists of a simple wire with the enable icon above it. This means the wire is only active when the enable signal (strobe in this case) exceeds a predefined threshold (double click on the enable block to set the threshold). When the enable signal (strobe in this case) is below the predefined threshold, nothing happens and the output holds its value.

  4. The decision subsystem is also enabled. This is because decisions should be made on every other interpolator output on average. The strobe indicates which interpolator outputs represent the desired signal space projections. An example of the enabled decision subsystem for binary PAM is shown below

Exercise

  1. Incorporate the carrier phase and symbol timing synchronization PLLs into a QPSK detector.

  2. For the detector input, use the From File block and set the Filename to qpskcompletedata.mat and the sample time to 1.

  3. Set the simulation parameters as follows:
    Simulation Time
    Start Time: 0.0
    Stop Time: 8*(2611+24)
    Solver Options
    Type: Fixed-step
    Solver: discrete (no continuous states)
    Fixed step size: auto
    Tasking and sample time options
    Periodic sample time constraint: Unconstrained
    Treat each discrete state as a separate task: Deselect

  4. Run the simulation.

  5. The detector produces approximately 2625 symbol decisions. (The exact number depends on the number of times the strobe is high. This, in turn, depends on the dynamics of your symbol timing PLL.)

  6. To find the data, look for the 16 bit SYNC pattern in the detector output. Once you find it, the following 3206 bit decisions correspond to 458 7-bit ASCII characters. Determine the message using either your Matlab script or an ASCII Table.

  7. Plot the filtered phase error (loop filter output) and use this plot to estimate how long (measured in symbols) it took for your carrier phase PLL to lock.

  8. Plot the filtered timing error (loop filter output) and use this plot to estimate how long (measured in symbols) it took for your timing PLL to lock.

  9. Plot the fractional interpolation interval (mu). What conclusions can you make based on this plot?
Brigham Young University - Provo | Fulton College of Engineering and Technology | The Church of Jesus Christ of Latter-day Saints
Department of Electrical and Computer Engineering, BYU, Provo, UT 84602 - (801)422-4012 - Copyright 2009. All Rights Reserved