Using Matlab + HackRF in Windows Environment
Use Hackrf in Matlab under Windows Environemnt.
Hackrf is always the best tool for playing or learning the SDR in my opinion. The common way of using it is in Linux environment, together with Octave for data processing. A much easier way of using it is just listening to the radio in Windows environment through all kinds of SDR softwares like SDRsharp. However, i really would like to find a way to use the hackrf more sophisticatedly in Windows environment. Luckily, there is a very good project on the github to support the hackrf development in Windows using Matlab & Simulink. There is also a video on YouTube describing how to install/complie this project https://www.youtube.com/watch?v=7dtikuo3BSw
But it seeme there is no longer any maintaince on this project, the issues about the installation or complie on different versions of dependencies or Matlab have not been answered anymore. So i spent a weekend to figure out how to correctly install it follwoing the video on the Youtube. After lots of failed attempts, i finally made it work. I record the whole procedures here and hope it could help those who are interested in. Hope u are enjoying
This project has very strong requirements on the version of the supporting softwares and dependencies. Even the Matlab version has to be R2016a. However, this Matlab version requirement is actually requiring the version of the MinGW64 adds on of the Matlab. When installation is done, we could switch back to a higher Matlab version to use it. And the speed of processing, from what i have seen, is actually better with the Matlab R2019a. It will not give OOOOO error report when using R2019a.
Installation Steps
1. Install drivers for HackRF
1.1 This step will not be discussed here. Once you have used SDRsharp in Windows, you must have installed this driver. Just download the Zadig then everthing is done.
2. Install Matlab R2016a
2.1 This step will not be discusses there. I have failed to use Matlab R2019a for installing this project. The root cause is the version of the Mingw complier of the Matlab. It seems the tdm64-gcc-4.9.2 is the only version which can successfully make this project in Matlab. I saw there are several issues on the installation using different versions of Matlab, so I strongly suggest you to use R2016a for installation. You can switch to a higer version to use it after installation. You are very welcome to debug why the make procedure will give errors when using higher version of Mingw (higher version of Matlab). Thanks in advance.
3. Install MinGW64 (tdm64-gcc-4.9.2) in Matlab
3.1 The current Matlab R2016a adds-on manager does not support automatically install MinGW64. You need to manually install it. Firstly we need to download the tdm64-gcc-4.9.2 from here https://sourceforge.net/projects/tdm-gcc/files/TDM-GCC%20Installer/Previous/1.1309.0/. Then we run the .exe file. 【Do not click the Check for updates files box】After installation, add the path to system environment. The Name is WM_MINGW64LOC and the _Value is D:\TDM-GCC-64 (Your own path). Finally execute mex-setup in Matlab and you should see following figure
4. Install MinGW64-posix-6.4.0
4.1 There are many blogs talking about how to install this, so we are not going to repeat it. The only thing you need to notice is that only the posix version is supported. I haven’t tried different version numbers, you are welcome to have a try and leave a message to me. If the online installation is not available for you, the offline installation is also ok.
5 Install cmake 3.6.2
5.1 The most important thing here is to use the correct version of the cmake. I haven’t tried another versions but in one blog it mentions to use the exactly version 3.6. So lets just follow it. It might be difficult for you to find this old version, i put all necessary softwares and dependencies in the end of this blog. You could also find it in this link https://download.csdn.net/download/icbm/9646285
6. Install cygwin64
6.1 There are lots of blogs on this topic, the only problem i meet is that I could not find the online mirror source on one of my computer. So i have to use offline installation. In case you have the same problem, i leave the link of the offline installation package in the end of this blog. You only need to install the basic cygwin64.
7. Download libusb
7.1 You can download the libusb directly on github https://github.com/libusb/libusb/releases . There is no requirement for the version. Just download and unzip it.
7.2 Copy the libusb.h from D:\libusb\include\libusb-1.0 (Change to your own path) to the include path of the complier, which is C:\Program Files\mingw64\x86_64-w64-mingw32\include (Change to your own path)
8. Download libhackrf
8.1 You can download the libhackrf directly from github https://github.com/greatscottgadgets/hackrf. We only need the libhackrf in hackrf-master -> host. Just download and unzip it.
8.2 Copy the whole libusb folder in previous step to the libhackrf folder.
9. Download simulink-hackrf-1.0
9.1 You can download it directly from github https://github.com/kit-cel/simulink-hackrf/releases/tag/v1.0. Just download and unzip it. Then create a folder named deps in it.
10. Run cmake
10.1 Open cmake_gui.exe in your cmake installation path.
10.2 Choose D:/libhacrf/src (Change it to your own path) for where is the source code. Choose D:/libhackrf/build (Change it to your own path) for where to build the binaries
10.3 Click Add Entrt, and manually add those entries as the figure shown below. If the Value is a path, select PATH for the Type, if the Value is a file, select FILEPATH for the Type, if the Value is a box, select BOOL for the Type and if the Value is empty, select PATH or STRING.
10.4 Click Configure and choose MINGW for complier. It will show Configure done if it is successfully executed.
10.5 Click Generate. It will show Generate done if it is successfully executed, as shown in the figure of step 10.3
11. Make libhackrf in cygwin64
11.1 Open the cygwin terminal and cd into the libhackrf/build. Then execute mingw32-make
11.2 Execute mingw32-make install in the same path. The figure below shows the resutls after these two steps. You can also find there are two folders named bin and include created in simulink-hackrf-1.0/deps
12. make in Matlab
12.1 Copy the hackrf.h from D:\simulink-hackrf-1.0\deps\include\Project (Changeb to your own path) to D:\simulink-hackrf-1.0\src (Change to your own path)
12.2 Open Matlab R2016a and select the simulink-hackrf-1.0 folder, run make command
12.3 Ideally it should successfully make. However, there might be an error as the figure shown below. To solve this, you could create an empty libhackrd.lib (right click -> create new .txt file, then change .txt to .lib) in D:\simulink-hackrf-1.0\deps\bin (Change to your own path). Then copy libhackrf.dll from the same path to D:\simulink-hackrf-1.0\build (Change to your own path). Finally add the build path into Matlab and make again.
12.4 Now if you see the figure shown as below, the make is successful.
13. Run Demo
13.1 Let’s have a look at the demo. Notice we need to add demos and build into the Matlab path. Plug in the HackRF and execute hackrf_find_devices.m to.fins the device. Then open the scope_demo in demos. After correctlt setting the sampling rate, gain, and center frequency, we could see the spectrum in Simulink. The figure below is the comparsion between spectrum from Airspy and Simulink.
13.2 BTW, after make in Matlab, we could use a higher version Maltab to run. Just add the demos and build into the Matlab path.
Link for installation package
Here I put all necessary softwares and dependencies in the link below (Except Matlab of course…).
transfer link: https://cowtransfer.com/s/2c0178772e934a or access [ cowtransfer.com ] and input extract code: 066ef3 to download;
—- The End —