Designing a Remote Sensor Network

Hi all. I am looking to implement a kilometer-scale sensor network for the observatory that I work for. I have a background in physics, but no knowledge of EE or networking. I was hoping you could help with some basic questions.

We are very interested in how seismic and acoustic vibrations impact our data. To investigate this, we have a collection of accelerometers and microphones permanently mounted to the observatory. I would also like to have a collection of temporary accelerometers and microphones which can relay their data in real time to our central building.

I am picturing a system where the raw +/-5 V signal from one or more of these temporary sensors is passed to an easily portable transponder. We are interested in data going up to 2048 Hz, so I think a sample rate of 4096 Hz is needed for the transponder. That transponder broadcasts the signal up to ~5 km via radio or Wifi (which one?). A receiver at the central location converts the wireless signal back to a voltage. This voltage signal can be integrated into the existing data system easily once it’s output as a voltage.

Is this feasible?


“Feasible” is a term approaching par with “fair” in terms of its elasticity…

One big question would be the timing constraints that apply to your measurements; if you just want to know whether the dirt in the back 40 was wiggling around dinnertime yesterday that’s one thing; it’d be quite another if you’re trying to locate the cow walking across it based on differential time of arrival. Most any system you’d come up with will introduce some latency, and depending on the communication scheme that latency might have a fair degree of variability. Not advantageous for acoustic bovine geolocation, that.

The fidelity with which you need to sample the data will also play a role, both in terms of data rate and probability of finding something approximating an off-the-shelf solution; crusty 8 or 10-bit data converters are a dime a dozen whereas 20+bit seismology-grade units are still a more rarefied breed.

Finally, the amount of effort, expertise, and expense you can bring to bear on the problem will put boundaries on the options; if you need a plug-and-play solution this weekend that costs $5 and has an instruction manual that consists only of illustrations that’s one thing, whereas if you’ve got $5M that needs burning and an electronics hardware engineering staff in the lab across the hall, that’d be another.

Offhand, I’m not immediately familiar with any sort of COTS wireless analog cable sort of system that’d seem suitable for your needs as I understand them presently. If you could help me better understand the specifics, I’d be happy to come up with some suggestions.