HDL OFDM Transmitter

This example uses:

This example shows an OFDM-based wireless transmitter implemented in Simulink® that is optimized for HDL code generation and hardware implementation.

This example shows the custom design of an orthogonal frequency-division multiplexing (OFDM) based transmitter. This transmitter model accepts payload data through the input port. It allows you to choose the modulation type and the punctured convolutional code rate of the data from a set of values. These two parameters control the effective data rate of transmission and they can be provided through the input ports of transmitter. The maximum data rate supported by the transmitter is 3 Mbps. The transmitter also accepts an input packetValid signal to control the transmission.

The transmitter in this example works in conjunction with the receiver in the HDL OFDM Receiver example. The transmitter has a MATLAB® floating point equivalent function described in the HDL OFDM MATLAB References example.

Transmitter Specification

This section explains the specifications of the transmitter related to the OFDM frame configuration and structure, bandwidth, and sample rate.

The transmitter model accepts two parameters, modTypeIndex and codeRateIndex, which allow you to specify the modulation type and punctured convolutional code rate, respectively, of the data. These two parameters are explained in the following tables:

modTypeIndex

    Value    Represents modulation type
    _____    __________________________

      0                BPSK            
      1                QPSK            
      2                16QAM           
      3                64QAM

codeRateIndex

    Value    Represents code rate
    _____    ____________________

      0              1/2         
      1              2/3         
      2              3/4         
      3              5/6

OFDM Frame Structure

Every OFDM system has a frame structure that shows the distribution of samples in the frequency domain across all its subcarriers. The frame structure is as shown in the figure. Each OFDM symbol is comprised of 72 subcarriers, and each OFDM frame consists of 36 OFDM symbols. The frame duration is 3 ms. The first OFDM symbol is formed by synchronization sequence (SS), second and third symbols are formed by reference signals (RS), and the fourth symbol is formed by Header. Data is filled from the fifth symbol to the last (36th) symbol. Pilots are inserted between data such that there is one pilot for every five data subcarriers as shown below. These pilots help detect and correct phase errors at the receiver.

The OFDM parameters used in the model are given below:

           Parameter              Value  
    ________________________    _________

    Sample rate                 1.92 Msps
    Subcarrier spacing          15 KHz   
    FFT Length                  128      
    Bandwidth of OFDM signal    1.4 MHz  
    Active Subcarriers          72       
    Left guard subcarriers      28       
    Right guard subcarriers     27       
    Cyclic Prefix length        32       
    Data symbols per frame      32       
    Pilots per data symbol      12

Model Architecture

The following figure shows the high-level architecture of the OFDM transmitter. There are five different signals that form the OFDM frame: SS, RS, Header, Pilots, and Data. SS, RS, and Pilots are the same for every frame. They are stored in seperate look up tables (LUT) and accessed whenever required. Header and Data vary based on the inputs given to the transmitter. Header bits are formed based on the modulation type and code rate input values. These header bits are processed through the Header chain as shown in the figure. Payload data is provided as an input to the transmitter. This data is processed through multiple stages in the Data chain. Individual stages in the Header and Data chains are explained in further sections.

These five signals are multiplexed based on their valid signals and stored in a RAM. The RAM holds these signals for a duration of one frame. Data stored in the RAM is read out and modulated by the OFDM Modulator block. The OFDM modulated signal is filtered with a passband frequency of 1.4 MHz and sent out as transmitter output.

File Structure

This example uses two Simulink models, an initialization script, and a MATLAB function:

whdlOFDMTransmitter.slx — This is the top-level model in this example. It has an OFDM Tx subsystem that refers to the whdlOFDMTx.slx model. There is external interface circuit for the OFDM Tx subsystem, which provides inputs and collects outputs from the subsystem. Running this model runs the remaining three files.

whdlexamples.OFDMTransmitterInit — This script initializes the whdlOFDMTransmitter.slx model. The script is called in the InitFcn callback of the model.

whdlOFDMTx.slx — This model implements the transmitter with total configurability. The whdlOFDMTransmitter.slx model refers to this model as explained above.

whdlexamples.OFDMTxParameters — A function that generates parameters required for the whdlOFDMTx.slx model. This function is called in the Model Workspace of the model.

Transmitter Interface

The top level whdlOFDMTransmitter.slx model shows the OFDM Tx subsystem and its interface.

Model Inputs:

modTypeIndex — Selects the type of symbol modulation to be applied to payload data, specified as a ufix2 scalar. This port accepts values 0, 1, 2, and 3 which correspond to modulation types BPSK, QPSK, 16QAM, and 64QAM.

codeRateIndex — Selects the code rate of punctured convolutional code to be applied to payload data, specified as a ufix2 scalar. This port accepts values 0, 1, 2, and 3 which correspond to code rates 1/2, 2/3, 3/4, and 5/6.

packetValid — Controls transmission of frames, specified as a Boolean scalar. If packetValid is 1, the transmitter outputs a valid OFDM frame for given input payload data, modTypeIndex and codeRateIndex values. If packetValid is 0, the transmitter outputs a dummy frame without the given input configuration and data.

dataIn — Input payload data, specified as a Boolean scalar.

All input ports run at a sample rate of 30.72 Msps to support different configurations.

Model Outputs:

txData — Transmitter output, returned as a complex scalar with fixdt(1,16,13) datatype sampled at 1.92 Msps.

txValid — Control signal that validates txData, returned as a Boolean scalar sampled at 1.92 Msps.

sampleInParam — Control signal used to sample modTypeIndex, codeRateIndex and packetValid inputs, specified as a Boolean scalar sampled at 30.72 Msps.

dataReady — Control signal that samples input payload data, dataIn, specified as a Boolean scalar sampled at 30.72 Msps.

Sample Input Signals

The Sample Input Signals subsystem samples modTypeIndex, codeRateIndex and packetValid signals based on the sampleInParam signal. This subsystem provides outputs only when there is an active sampleInParam signal and retains the previous outputs if the sampleInParam signal is inactive. The subsystem stores given input modOrder and codeRateIndex values in two LUTs as shown in the figure. If packetValid is 1, corresponding modOrder and codeRateIndex values are selected from LUTs. The Source Coding function maps modOrder values 2, 4, 16, and 64 to the corresponding modTypeIndex values 0, 1, 2, and 3 and sets 1 (QPSK) as the default value. It also takes codeRateIndex values 0, 1, 2, and 3 from the LUT and sets 0 (1/2 code rate) as the default value. If packetValid is 0, the LUTs are ignored.

Select Payload Data

The Select Payload Data subsystem selects input payload data based on the dataReady signal. It has a counter that increments when the dataReady signal turns active. There is an LUT that stores the given input payload data. Data from this LUT is selected based on the counter value. There is an additional circuit that returns zeros as data after the counter reaches maximum count.

Structure of the Transmitter

The whdlOFDMTx.slx model is called within the OFDM Tx subsystem. It generates an OFDM transmitter waveform by processing input signals in multiple stages as shown below.

whdlOFDMTx

Frame Controller and Input Sampler

The Frame Controller and Input Sampler subsystem has a counter that increments for a frame duration of 3 ms and then restarts. The counter restarts from 0 at the beginning of every frame and generates the sampleInParam signal to sample input signals modTypeIndex, codeRateIndex, and packetValid. Based on the sampled modTypeIndex and codeRateIndex values, corresponding transport block size for the frame is selected from the Transport Block Size LUT. Then, the dataReady signal becomes active to accept the input data, dataIn, of size equal to transport block length. If sampled packetValid signal is false, dataReady remains inactive indicating generation of a dummy frame that does not require any input data. This subsystem also generates control signals for SS, RS, Header, Pilot, and Data generation that is carried out in further stages.

Frame Generator

The Frame Generator subsystem consists of five subsystems that generate SS, RS, Header, Pilot, and Data signals, which are later OFDM-modulated.

Frame Generator/SS

The SS (Synchronization Sequence) subsystem accepts ssSet control signal, generated from the Frame Controller and Input Sampler subsystem. It is generated considering the length of SS sequence. This signal initially resets the counter. Then, the counter keeps incrementing and simultaneously returns SS from an LUT. Once ssSet becomes inactive, the counter stops. Output from LUT is upsampled by a factor of 2 to maintain the same sample time as that of the Header and Data subsystems. RS and Pilot subsystems operate in a similar way by storing the sequences in LUTs and accessing them whenever required.

Frame Generator/Header

The Header subsystem accepts modTypeIndex, codeRateIndex and fftLenIndex as inputs. A headerSet signal starts the header formation. The Header Formation function converts the modTypeIndex and codeRateIndex values into their binary equivalents. For example, a modTypeIndex value of 1 is converted into two bits 01. Similarly, codeRateIndex values are converted into two equivalent bits. To learn more about these indices, refer to Transmitter Specification. fftLenIndex is not configurable and its value is fixed to 0. It is converted to 000, which represents an FFT length of 128. fftLenIndex, modTypeIndex, and codeRateIndex are represented using 3, 2, and 2 bits, forming a total of 7 bits. Additionally, 15 spare bits are added, all currently set to 0, forming a total of 22 Header bits.

These 22 bits are processed as shown below. For proper error detection, 8 CRC bits are padded using the General CRC Generator HDL Optimized block with [8 7 4 3 1 0] as the CRC polynomial. These 30 bits, that is (22 + 8), are encoded using the Convolutional Encoder block with [171 133] as the polynomial and a constraint length of 7. The encoding is done in terminated mode which adds 6 null bits, that is (7 - 1), to the CRC padded data. After encoding, these 36 bits result in 72 bits due to the 1/2 rate encoding. The output of the Convolutional Encoder block is a two-element vector that is serialized using the Serializer1D (HDL Coder) block leading to rate transition by a factor 2. These 72 bits are BPSK-modulated using the LTE Symbol Modulator block to form Header symbol.

Frame Generator/Data

The Data subsystem stores input payload data, dataIn, and processes it through the Data chain.

Frame Generator/Data/Data and Control Signal Generation

The Data and Control Signal Generation subsystem consists of a RAM where input payload data, dataIn, is stored. A dataSet signal reads data from this RAM. This subsystem generates start, end, and valid control signals for the RAM data. It even selects the puncture vector based on the codeRateIndex.

Frame Generator/Data/Data Chain

Payload data from RAM and the control signals are given as inputs to the General CRC Generator HDL Optimized block, which appends a 32-bit CRC to the data with [32 26 23 22 16 12 11 10 8 7 5 4 2 1 0] as the CRC polynomial. This CRC-padded data is scrambled with $x^7+x^4+1$ as the polynomial and [1 0 1 1 1 0 1] as the initial state. The scrambled data is encoded using the Convolutional Encoder block in terminated mode using [171 133] as the polynomial and a constraint length of 7 . This output is punctured using the Puncturer block with the puncture vector selected in the Data and Control Signal Generation subsystem. The output of the Puncturer block is a two-element vector and is serialized using Serializer1D (HDL Coder) block. The resultant data is modulated using the LTE Symbol Modulator block using the modulation pattern selected based on modTypeIndex.

Multiplexer

The Multiplexer subsystem multiplexes the SS, RS, Header, Pilot, and Data signals based on the corresponding valid signals that are generated by the Frame Generator subsystem.

Frame Formation and OFDM Modulation

The Frame Formation and OFDM Modulation subsystem takes in the multiplexed dataOut and validOut signals and writes them into a Dual Rate Dual Port RAM (HDL Coder). This RAM reads and writes data at different rates. Data is written into the RAM at 61.44 Msps. An LUT contains addresses for writing data into RAM. The RAM is filled with data such that it forms an OFDM frame structure as shown in the previous sections.

The OFDM Modulator Valid Generation subsystem generates a valid input signal for the OFDM Modulator block at a sample rate of 1.92 Msps. This valid signal is in synchronization with the ready signal of OFDM Modulator. The valid signal is also used to generate a RAM address to read data from the RAM. The Control Packet Output subsystem selects the OFDM Modulator output based on packetValid. It gives out valid OFDM output in the presence of packetValid and a dummy frame with random SS in its absence.

Discrete FIR Filter HDL Optimized

The output of the Control Packet Output subsystem is filtered with a passband frequency of 1.4 MHz using the Discrete FIR Filter HDL Optimized (DSP System Toolbox) block . The filter coefficients are computed by the whdlexamples.OFDMTxParameters function. Output of the filter forms the final output of transmitter.

Run the Transmitter

The transmitter can be connected back-to-back with the receiver that is explained in the HDL OFDM Receiver example. For more information on how to use the transmitter and receiver Simulink models back-to-back, refer to the HDL OFDM MATLAB References example.

To run the transmitter model, OFDMTxVerification.m script is provided with this example. The script chooses a custom frame configuration, and payload data, and simulates the model. The script also collects the simulation outputs and validates them.

NOTE: These files are not available on the MATLAB search path. To copy these files locally to the user path, you must open this example.

Verification and Results

In this section, the OFDM Transmitter Simulink model is validated by comparing its output with its floating point equivalent function, whdlexamples.OFDMTx. For more information on this MATLAB function, see the HDL OFDM MATLAB References example. To compare the output of the Simulink model with the MATLAB function, run the OFDMTxVerification.m script. Note that for comparison with whdlexamples.OFDMTx, packetValid should be continuous.

>> OFDMTxVerification

### Starting serial model reference simulation build
### Model reference simulation target for whdlOFDMTx is up to date.

Build Summary

0 of 1 models built (1 models already up to date)
Build duration: 0h 1m 36.745s

HDL Code Generation

To generate HDL code for this example, you must have HDL Coder™. Use makehdl and makehdltb commands to generate HDL code and HDL testbench for the OFDMTx subsystem. Testbench generation time depends on the simulation time.

The resulting HDL code is synthesized for the Xilinx® Zynq®-7000 ZC706 evaluation board. The post place and route resource utilization is shown in the table below. Maximum frequency of operation is 227 MHz.

       Resources       Usage
    _______________    _____

    Slice Registers    5819 
    Slice LUT          3631 
    RAMB36             5    
    RAMB18             12   
    DSP48              24