My 2c to the discussion. This is not mechanic specific but more food for thought.
If I may I'll begin with an oversimplified rant (and I do mean it, but there's a lot going on behind this reasoning) of how most RL military longer-range (as in stuff viable for space) sensors work.
They're all based around particle detection, and emission but that's self explanatory (mostly photons of varying wavelengths).
I won't be going into other types of sci-fi sensors since then it's more literature and preference than anything else.
Range and resolution is a combination of:
Wavelength/Bandwidth: Radio<->MW<->IR/heat<->Visible Light<->UV (in modern day practical solutions) but cost, power and size increases exponentially with output and bandwidth.
Aperture/antenna-size: This can be actual massive antennas, networked arrays or synthetic arrays (ie the sensor is moving forming a virtual large-size antenna over time and processing for LR detection with multiple returns).
Output Power: the more the merrier but it's like shining a flashlight in a dark forest, your beam will be seen at much longer ranges than the range it provides in any case.
Processing power: Think SETI@home or supercomputing if you are to make the most of sent/collected information.
High resolution can, for explanatory reasons, be simplified down to a function of wavelength, time, distance, antenna-size and processing-power.
Long wavelength radio provides less possible resolution and requires larger antennas (actual or arrays) compared to high wavelengths and for single non-networked sensors the wavelength provides a fixed range-vs-resolution (ie arc second limitations vs array size) but can get away with simpler antennas/apertures.
High wavelength does provide higher bandwidth and good resolution but does require a lot of power to be useful (basically techniques for various degrees of collimation/beam-forming is the only thing making it feasible for active sensors over range and then with limited FOV (or time to scan a section of space if you will), think laser rangefinding). There's always a problem with multipathing from various objects the higher you go (basically increases with wavelength/resolution) which means you'll have to rely on encoding your emissions more which requires more processing-power and so forth.
We could dig deeper into beamwidths and Fresnel sizes etc but it's sufficient to say that narrowbeam is better than omnidirectional in terms of power req vs resolution vs overall noise.
Now one might think that I'll build a gazillion of smaller sensors in all the different flavors and spread them out everywhere which is fine in principle but less so in practicality (power, maintenance etc) .
If they're active and share bandwidth/wavelength they need to be completely synchronized or you'll basically have created a huge noise-machine which could be further diminished in use by a hostile adding to the noise-level (ECM) so they'd all need to be networked and then it's just a big can of worms in terms of micro-management (as far as a game goes).
Passive detection is important but, for the sake of simplicity, should give much less accurate positions (a single passive sensor or multiple non-networked ones would in reality only give a very rough estimate from bearings as they can't range the target). Large networked arrays of passive sensors could be modeled as single installations of great size instead (and cost accordingly). For a passive sensor to be really useful it needs to be networked and convey collected information in one way or another which means that even a passive sensor is actually active (in terms of transmissions for networking). Emissions control is paramount to avoid passive detection. That includes getting rid of heat in various clever ways while absorbing as much incoming photons as possible (man made or natural) or deflecting them in directions where the enemy has no sensors.
To keep it simple I would suggest a noise-level for a given system. If you add a gazillion of sensors with all the bells and whistles and large output-power you will get diminishing returns somewhere along the line where you're basically just adding background static noise reducing the efficiency of your active sensors and broadcasting a huge bulls-eye for passive sensors.
Background ECM could add to said noise-level and so could environmental stuff (nebulae, quasars etc) with the base-line that it's always easier to destroy than to create.
TLDR; (without any further regard to other mechanics)
Sensor basic attributes could be divided into power, bandwidth, cost and size.
Complex systems with large BW would be expensive and have very high maintenance requirements.
Systems with a large power output could be a bit cheaper (but with higher power requirements) but then visible from afar and noisy.
Large systems are. . . large and thus cost accordingly and require processing power and networking (if arrayed). Large stuff is easier to hit and harder to protect.
Lots of active sensors with lots of power means good coverage but also lots of noise (reducing resolution at range).
Also. Atmospheres are bad. Earths atmosphere for example is semi transparent to visible light and some ELF but practically opaque to a lot of other wavelengths (as far as reasonable deep-space sensors go). Sensors on planets can be built to huge sizes and the orbital movement can be used for very long range synthetic arrays but it's highly dependent on the atmospheric conditions of the specific planet.
Also. Anti-radiation passive warheads are super simple compared to active sensors and kills expensive active sensors and comms IRL. Like Alex wrote they're the scourge of active sensors and really really nasty IRL. In sci-fi terms in space a ARM could creep along with not a hurry in the world (loitering mode) and then jump you when you turn on your sensor back on.
Real ARMs guide by back/side lobe artifacts from the transmitter so they don't even have to be in the main sequence lobe (and can thus come from anywhere).