OH2BNF’s Plan for a Large Scale Raspberry SDR (LSR-SDR) Based on RTL-SDR Dongles

Thanks to OH2BNF for writing in and sharing his plan to build a "Large Scale Raspberry SDR" (LSR-SDR), which will be based on RTL-SDR dongles. To create the LSR-SDR he plans to take a 19" rack which can support up to 40 Raspberry Pi 3's, plus up to 160 USB devices, and turn it into a massive SDR array. The rack is key as it allows for simple power management of all the Pi's and other devices to be connected.

OH2BNF plans to connect 20 or so RTL-SDRs, with some operating individually and with others operating coherently via a common external oscillator. The rack may also contain some transceivers, an ICOM IC-7300, antenna switches, upconverters, LNAs and other hardware too. Once completed he hopes to move the system to a low RFI environment and operate the unit entirely remotely. With this he hopes to solve his local RFI issues. He also writes regarding applications:

Primary objectives are to incorporate automated adaptivity to the system at large – for example leveraging on band condition information, WSPR (Weak Signal Propagation Report) & friends, automated signal detection and decoding, great flexibility in terms of individual cluster nodes being able to fast respond to various needs and tasks, strong emphasis in parallel processing where applicable depending on the problem type and dataset, support for multiple end users benefiting from the computing and reception capacity of the cluster – to name the most significant.

It's an interesting idea for sure, and we hope to see some updates from OH2BNF in the future.

The Raspberry Pi 19" Rack
The Raspberry Pi 19" Rack
Subscribe
Notify of
guest

13 Comments
Inline Feedbacks
View all comments
EmmanuelF

Beside the RF noise,
Why choosing the worst ever found in the universe USB performing SBC to build a cluster of … USB SRD receivers ???

OH2BNF

Dear EmmanuelF,

if you read the actual blog you will find all the answers, including most notably the fact that this is a lot of fun, incorporates several items and that I am in general very happy with both the Raspberries as well as v3 dongles. Looking at my table, I also find that my glass is half full of water.

If you don’t mind, I would be delighted to hear about your cluster. As with all projects, it is always beneficial to compare experiences and try to contribute to the community at large.

Tom

Forgive my ignorance, but what is the benefit of using an rpi cluster? The SDR is basically just pipes RF into USB, so the choice of computer should be irrelevant, not withstanding RF isolation issues. The cost of 40 units is in the order of about $200-$400. The power, switching, infrastructure and chassis is probably another $200.

Is there any real benefit to this over a more capable, low power Avoton with 16 cores, or is it just academic?

OH2BNF

Hi,

thanks for asking. This is a multifaceted project which involves examination of a large number of items, amongst them Raspberries. I did consider for a fairly long time using Odroid SBC’s, but had such a lot of fun already with Raspberries, that I ended up using them.

There are after all 160 cores; the key is to programmatically deal with a distributed computing setup, not necessarily “how fast” it is (it is fast enough however for the tasks). If overall concept is proven satisfactory, nothing prevents investing in different type of hardware should the performance become an issue.

It has been apparently quite not evident for most that this is most of all a man-size sandbox; very early on, I looked at some of the very expensive solutions being sold for military sigint. Those cost easily on the order of magnitudes more. I wanted to see what can be done using common off-the-shelf parts; that is one of the motivational elements behind the project.

I would actually turn the response topsy-turvy: in many – but not all – aspects, it *is* an academic study. However, I have been quite happy with Raspberry and SDR so far in smaller settings, and anticipate to be more so with a large number of them.

OH2BNF

Hi,

the bandwidth available from devices other than v3 is not at all the issue. In fact this project has dimensions and objectives that extend far beyond just the bandwidth question; and perhaps it needs to be clarified that the intention is specifically to use 40 Raspberry devices, not to limit the number of them.

As stated we are still in the first – infrastructural – phase, and we’ll get to the interesting stuff later on. Certainly as has been indicated there are some interesting possibilities forecasted. Setting up a complete, remote manageable and robust HPC cluster – despite the admittedly low computing performance of the Raspberries – nevertheless takes time, in particular the PXE booting is nowhere near robust enough at the moment.

Broadening the bandwidth of reception seems to turn up frequently in discussions, but is really nowhere near the top requirements for the operation. Stuff of interest rather revolves around the cluster at large, its adaptivity to varying conditions, pattern detection and automation.

Parallel processing when applicable and supported by the “problem” and dataset also is one of the priorities. For the vast number of radio related software, examining the possibilities and testing, testing, testing 🙂 is what I am most looking forward to.

Seasalt

I tried to reply on your Blog but I could not find a way to comment. I run “NOSCRIPT” addblocker in Firefox so it may have blocked comments.

Can your 40 Raspberry Pi Cluster run a program like GQRX in a more powerful and useful way etc?

Please keep us informed as I think your idea is the future for a lot of HAM radio computer buffs.

What is the name of the Raspberry Pi Cluster software?
Low cost powerful experiment computer for Radio experiments.

PS I would be interested if your Rasp Pi Cluster could be useful for Passive radar decoding.

OH2BNF

There is a large amount of (radio related) software that may benefit from a distributed computing setup. Not necessarily so with the GQRX, but potential exists elsewhere.

The clustering solution is almost certainly going to be SLURM https://en.wikipedia.org/wiki/Slurm_Workload_Manager but haven’t gotten all that far yet. SLURM is used – according to the wiki page – in more than 60% of the world’s supercomputers, so I am sure it will be good enough for my little one. I have past experience with SGE and LSF, so the basic principles are something I am familiar with.

I am sure there will be literally zillions of use cases for the cluster, once the solution has been ironed out, a proper QTH found and most important – the antennas…

Seasalt

I want to here more about this. Particularly the Cluster software he is using. Is the available Raspberry Pi software very user friendly. For things like GQRX would a cluster give you more COMBINED bandwidth. etc?

Ben Barrett

Would be interesting to see what signal-analysis capabilities could be attained using OpenCL on the RPI GPU’s, esp with that many nodes!

andrej

lol how to deal with the noise?
When I was used an Pi and compare the noise between Pc my Pc wins.

SA5KPR

If he want to cover a larger continuous spectrum, it might be a better deal to get HackRF clones. The bandwidth/$ should be about the same for an HackRF clone and quality RTL-SDR… The real saving is if he need a lot fewer Pi’s.

Philip Hahn

Or a LimeSDR – 60MHz bandwidth over USB3!

SA5KPR

Yeah, If that works well you could cover basically all HF HAM bands + 6 m with a single SDR, you probably need around 9 RTL to do the same. But you also need SBC’s with USB3, they exist but it’s not as widespread yet.