Hacker News Reader

Comment by OldSchool | Hacker News Reader

Hmmm, I was kind of hoping for an explanation of how the control system determines how to modify the highly directional antenna's orientation to continue to maximally "illuminate" the target once it's been selected from a general scan.

Can anyone describe the solution to this?

Example idea: Once selected for "lock," the antenna scans in small circles and migrates its central point toward the point on the scanned circle that provides the strongest return, allowing for variation due to noise sources.

View on HN

There is a book on it[1]. But suffice it to say that the radar is computing an "exit vector" when the target it moving away from center and applying a correction to the pointer. I built a similar system with LEDs when building a tracking system for two moving robots. In my case I used a parabolic dish (a solar cigarette lighter) and a line of LEDs (bar graph display) as detectors rather than lights. Since the target robot and the pursuing robot had their beacon and search dish in the same plane I could reduce the problem to a managable bit of 2D geometry. As the signal went off axis you could turn the robot to re-align by applying the opposite rotation.

[1] "Multiple Target Tracking with Radar Applications" - Sam Blackman (http://books.google.com/books?id=Ag9TAAAAMAAJ&q=tracking+mul...)

Thanks for the thoughtful reply and reference!

Are you asking if the antenna gets moved to illuminate the target once acquired? I apologize if I misunderstood. If that's the case, the antenna is a solid state phased array; no movement is required. Mobile communications (cellular and satellite) spot beams work under the same principal.

http://www.northropgrumman.com/Capabilities/MESA/Documents/a...

Of course some radars did physically move their dishes to stay locked on to a target. I'm in the middle of The Invention That Changed the World (spoiler alert: it was radar) and I was stunned to learn how well this worked with analog circuits. An SCR-584, paired with an M9 analog computer, could detect that a plane was moving out of the center of its conical scan, and then convert that into a signal to the motors that drive the dish (and its associated antiaircraft guns).

http://en.wikipedia.org/wiki/Conical_scanning

http://en.wikipedia.org/wiki/SCR-584_radar

Thanks for pointing me to that book - it sounds like something I'd read.

Yeah phased arrays have probably been the rule for decades at that price-no-object level.

Even at the consumer level, there have been phased-array marine -sonars- (e.g. Interphase brand product line) for at least ten years for "only" 4-digit prices. Spatial resolution is of course tied to the number of elements but they do present more information that your typical 1-d "fish finder."

Reply on news.ycombinator.com