Re: [Discuss-gnuradio] USRP Structure

[mleech]

Since most SDRs out there have fully reconfigurable-by-the-end-user FPGA and firmware images, I don't think the notion of "compromise"
 has much meaning in this context, further because access to the devices is freely available to ordinary user-level processes, they can ask the radio to do whatever they want.

Most SDRs that we discuss here are used in R&D, and only a very few in "services" where type-acceptance is required.  Presumably, in the fullness of time, getting type acceptance would require the integrator to demonstrate some kind of "protection" for the radio.  But SDRs as we know them here are just "dumb" components.    It's a bit like asking a mixer or RF amplifier or synthesizer to "tamper-proof itself".

My personal opinion is that asking general-purpose hardware to enforce some arbitrary notion of regulatory compliance in this area is silly, unproductive, and ultimately doomed to failure, quite apart from the wide-reaching implications for the industry in general.

My "day job" is at a company where we "tamper proof" software on general-purpose computers at the behest of the Media Industry. It amounts to building perpetual-motion machines--it cannot be done in the strictest theoretical sense.  In a practical sense, you can keep the
casually-curious out of your "stack", but you cannot protect against the determined--they have infinite access to the hardware and software, and will eventually find a way around any "safeguards" you put in place.  So, in the first instance, the "lockdown" software is utterly unnecessary, and in the second instance, it is woefully inadequate...

 

 

 

On 2015-09-09 14:51, Logan Wu wrote:

Hello,

Recently I read a paper on cognitive radio security (Secure
reconfiguration of software-defined radio). It highlights that the
operating system of cognitive radio node may be compromised as the
malware can exploit software vulnerabilities. I am wondering if the FPGA
and firmware are part of the OS? And can they be compromised during
runtime by malware?

Thank you,
Logan