wizard1073
Posts: 23
Joined: 9/27/2013 Status: offline
|
Okay, got a really good chapter 1 in the OTH bible. It's the only reference I could find that covers scan rates in any detail. I'm a microwave radar engineer, so OTH requires a bit of bridging of my knowledge. If someone more knowledgeable shows up and can comment, please do! The OTH bible is _High Frequency Over the Horizon Radar: Fundamental Principles, Signal Processing, and Practical Applications_ by Guiseppe Aureliano Fabrizio. I think he works for the Australian MoD. My second source is Skolnik's Radar Handbook, 3rd Edition, Chapter 20 "HF Over-the-Horizon Radar" by James M. Headrick (retired from the Naval Research Lab) and Stuart J. Anderson (like Mr. Fabrizio, also from the Australian Defence Science and Technology Organisation). Since CMANO is a capabilities-based simulator, I am assuming that you don't want to make an OTH radar model. It's hard to recommend a correct scan timing as a result, but there might be a way. Consider the following information: OTH skywave radars typically have multiple modes: 1) Air route monitoring 2) Barrier surveillance 3) Strategic waterway monitoring 4) Ballistic missile launch detection 5) Remote sensing and hurricane tracking The modern OTH radars we use for target cueing will typically do air and surface monitoring (1 and (2 or 3)), but there is a time and space tradeoff. Typically they will either cut the search sector down, or they will look for surface in one area and air in another. If they do both in the same search sector, they will have to interleave the modes. The air modes will get used a lot more often, because air targets need more frequent updates to maintain track accuracy. Search sectors are typically broken into Dwell Illumination (or Interrogation) Regions (DIR). Basically, this is where you point the transmit beam and hold for enough time to get a decent signal-to-interference ratio. Multiple receive beams will be formed in a fan to search multiple azimuths within the DIR. A typical DIR might be 300km wide (at range) due to a transmit azimuth beamwidth of 8-12 degrees, and be composed of perhaps 20 (or more) receive beams (0.2-2.0 degrees azimuth beamwidth). If you also shift frequency to move the beam to get more rings of DIRs (which are 500-1000km long), then you might be looking at 40 or more DIRs. Within a DIR, the radar will dwell for some length of time, in order to develop a good signal. For air targets with good resolvable Doppler returns (which separate them easily from the surface clutter), the Coherent Integration Time (CIT) is 1-4 seconds. For surface targets with low resolvable Doppler return (because they move just a little faster than the surface clutter), the CIT is now 10-40 seconds or longer (perhaps 100+ seconds if the ionosphere is not cooperating, or the target is not moving much radially with respect to the radar). So then, for air mode, for a 90-degree search azimuth, we might have 40 beams leading to a search frame time of 40 to 160 seconds total. For surface mode, we might have 40 beams leading to a search frame time of 400 to 1600 or more seconds. If interleaving both and trying to do the full search sector with no tradeoffs, there will be some multiple of air search frame times relative to the surface search frame times. If the ratio is m to 1, then the total frame time is [m*(air search frame time) + (surface search frame time)]. The value of m is based on the persistence requirements. Radial range resolution of the radar might be somewhere between 3 and 30km, depending on the Pulse Repetition Frequency. For ship detection, it might be 5-10km uncertainty. For air detection, it might be 15-20km uncertainty. In cross-range, the resolution is tied to the beamwidth (unless there is some interferometric or monopulse-like processing occurring), so you might see 10-50km uncertainty. Localization is another problem because the ionosphere is dynamic. You don't quite know exactly where the ionospheric reflection occurred, and at what angles it deflected. As a result, you can only say with confidence that you placed the beam within 10-20km of where you think you did. To get around that, you can put down known reference points, which either have a simultaneous and identifiable radar return (like an island or structure that can be separated from the clutter), or have transponders that identify themselves and their location, in which case you get localization accuracy of less than 5km. Once you have a detection, the operator might be able to revisit more often to maintain detection. With military OTH radars, I think that is a safe assumption. How often to revisit leads to yet another tradeoff. Now, you will rarely wait the full frame time to develop a new detection. When we do radar modeling, we typically randomize uniformly within the frame time, to account for "the radar was pointed at X when you entered the search sector at Y, but it only took (frame-time minus random-time) to catch up." In conclusion, then, the frame time should be somewhere between 40 and 1600 seconds. (Aren't you glad you asked? ) Seriously, though, if you can differentiate between air and surface targets, then you might be able to tailor it to [40-160] seconds for air and [400-1600] seconds for surface, and randomize (or divide by two for the average) to make the delay closer to what is expected in real life. Alternatively, add a user- (or designer-) selectable mode for the radar (air vs surface vs both). Let me know if you have more questions. I enjoy finding these kinds of answers.
< Message edited by wizard1073 -- 6/1/2016 2:48:50 AM >
|