Multichannel FIR Filter for FPGA
This example shows how to implement a discrete FIR filter with multiple input data streams for hardware.
In many DSP applications, multiple data streams are filtered by the same filter. The straightforward solution is to implement a separate filter for each channel. You can create a more area-efficient structure by sharing one filter implementation across multiple channels. The resulting hardware requires a faster clock rate compared to the clock rate used for a single channel filter.
Model Multichannel FIR Filter
modelname = 'dspmultichannelhdl'; open_system(modelname);
The model contains a two-channel FIR filter. The input data vector includes two streams of sinusoidal signal with different frequencies. The input data streams are processed by a lowpass filter whose coefficients are specified by the Model Properties
InitFcn Callback function.
Select a fully parallel architecture for the Discrete FIR Filter block, and enable resource sharing across multiple channels.
systemname = [modelname '/Multichannel FIR Filter']; blockname = [systemname '/Discrete FIR Filter']; set_param(blockname,'FilterStructure','Direct form symmetric'); hdlset_param(blockname,'Architecture','Fully Parallel'); hdlset_param(blockname,'ChannelSharing','On');
You can alternatively specify these settings on the HDL Block Properties menu, which you access by right-clicking a block and selecting HDL Code > HDL Block Properties.
Run the example model and open the scope to compare the two data streams.
sim(modelname); open_system([modelname '/Scope']);
Generate HDL Code and Test Bench
You must have an HDL Coder™ license to generate HDL code for this example model. Use this command to generate HDL code for the Multichannel FIR Filter subsystem. Enable the resource use report.
Use this command to generate a test bench that compares the HDL simulation results with the Simulink model results.
Compare Resource Utilization
To compare resource use with and without sharing, you can disable sharing resources across channels and generate HDL code again, then compare the resource use reports.