Heh heh - processing in the frequency domain makes a lot of this easy. The chief issue is the current FFT resolution compared the bandwidth you want. The "cheap, quick, and dirty" solution is sum the power in all the bins within the bandwidth you want. I suspect using some shaping on that would add a decimal place to the accuracy as well as more processing demand on the system.

Now, if you use receivers as the filtering mechanism running a reverse FFT on that bandwidth (and a little bit more which will be super close to zero power and gives a power of two number of FFT steps) gives you time domain I and Q signals. And, I wish SDRC did this so its S in its SINAD fiddling would not be spoofable, square I, square Q, sum them together, and take 10 log that power equivalent. That gives the continuum measurement as well as a very good S for calculating SINAD ratios. (As Simon curls up in a corner screaming "I didn't read that! I don't wanna do it!")

(Snarky today Joanne? Yeah, just had the telephone guy come to deal with noisy phone lines when it gets windy. It was still air. So, in another half hour the internet vanishes here as the afternoon winds kick up. This time it's war. A couple days ago the noise made a telephone consult with my doctor a miserable series of "say again" and CRACKLE! CRACKLE! CRACKLE!)

{^_^}