Title: From Fixed Scan Scheduler (FSS) to Dynamic Scan Scheduler (DSS)
1ANT-Based Dynamic Scan SchedulingANTSNortheaster
n/Sanders
2Outline
- The scheduling problem
- Use Case Fixed Scan Scheduler
- From FSS to Dynamic Schan Sched. (DSS)
- Dynamic Scheduling via negotiation
- Plans
3Scenario
A
B
C
4Scenario - discussion
Consider a threat pulsed radar in surveillance
mode with pulse-repetition interval ? in the
order of 1 msec (200-km unambiguous range). (The
interval ? can be as small as 10 ?s.) The
principal lobe of its antenna pattern is a fan
beam in elevation with a 1-deg azimuth beamwidth.
The beam scans 360 deg in 3 sec. Therefore, its
illumination time at a point fixed in the far
field will be 8 msec. (3sec/360) The desired
EW-receiver revisit time, , is therefore 8
msec. This is the largest time interval that
guarantees that the illumination of the threat
can be captured by the EW receiver (the
receivers dwell time and the emitters
illumination time intervals will overlap). At
least three pulses must be detected in order for
the EW receiver to correctly identify the
emitter. (Parameter M on next page M gt3.)
Therefore, the EW-receiver dwell ? cannot be less
than 3 msec. This assumes that the Pulse
Repetition Interval (PRI) is 1msec. If PRI is
smaller than 1msec, then the dwell time can be
shorter.
5Emitter Parameters
Emitter Signal
Time
Pulsewidth
Pulse-repetition interval
Illumination time
Emitter revisit time
Desired EW revisit time Emitter illumination
time
Desired EW dwell M emitter PRI (M is an
integer)
6 Scenario-numbers
- Number of emitters 2000
- Emitters to track 50-200
- Average dwell time 3-30 msec
- Revisits per second 2
- If tracking 50 emitters, using 10mSec dwell times
each, how many revisits per Sec? - 1,000 ms/(50 x 10ms) 2 max.
7Fixed Scan Scheduler (FSS)
Definition Using prior knowledge, mission
planners construct a Static, Mission-Specific
Scan Schedule or a Fixed Scan Schedule off-line
for use during future missions.
8Fixed Scan Schedule Construction
9FSS Example
Schedule
A1 B1 A2 B2 A3 C1 A4
10Parameters - continued
Assume a K number of EW receivers are provided,
each covering an instantaneous bandwidth ?f.
Assume the spectral region of surveillance covers
a frequency range from f1 to f2.
Therefore, the surveillance coverage consists of
L number of bands any K of which can be covered
instantaneously
If
then we have a problem (cannot cover all)
A threat list of emitters of interest shows that
they operate in only L of the L frequency bands
from f1 to f2.
11Parameters -cont.
Frequency band l of the L bands contains an Ml
number of emitters, each having a desired
revisit time Tlm and desired dwell ?lm , m 1,
2, ???, Ml .
f1
f2
If ?lm gt ?ln , then whenever the EW system
revisits emitter m, it will also revisit emitter
n, provided that revisit time is Tlm and it is
the smallest In other words Trevisit min
Tlm ,Tln ?dwell max ?lm , ?ln
12Constraints on parameters
Tn ? EW revisit time for nth emitter.
?n ? EW dwell time for nth emitter.
Consequence is degradation of probability of
intercept.
13What follows
- There are two aspects to scheduling receivers
- performance of a receiver (Measure of
Effectiveness (MOE), when and for how long) and - the value of that performace to the system
(Figure of Merit (FOM), how important is that
threat) - The following slides focus on MOEs for receivers
14Metric of Scan-Schedule Performance
Example 1 (one pulse)
Rn ? Event that EW receiver is ready to receive
transmission from nth emitter when transmission
occurs.
In ? Event that EW receiver intercepts
transmission from nth emitter.
15Example 2 (three pulses)
R ? Event that the EW receiver is ready to
receive emitter transmission when it occurs.
p ? Conditional probability of EW receiver
detecting a pulse when it occurs, given that the
EW receiver is queued to receive it.
I ? Event that EW receiver detects at least three
of the emitter pulses.
Greatest integer in x.
16Example 3 (Mn pulses)
Minimum number of pulses intercepted that is
required to perform task (detection, emitter
identification, direction finding).
Suppose EW receiver dwells long enough for
emitter to transmit Mn pulses.
pn ? Probability of detecting a pulse from nth
emitter.
The longer the dwell ?n, the larger Mn and the
the higher PIn.
17MOEs
- Now we focus on MOEs for the Scheduler
- The following slides need to be reviewed
18Resources, Tasks and Constraints
19Two scenarios - two sets of MOEs
MISSION
Egress
Ingress
Must survive to complete mission destroying
target(s).
Must survive to get home intact.
20INGRESS (Surviving to hunt/destroy targets)
EW Search (Searching to protect against lethal
threats)
RESOURCE
Required Reaction Time to Survive
Probability of Surviving, given that threat is
encountered P(S?)
Probability of encountering threat P(?)
Dwell Time
Estimated Lethality
Revisit Time
Threats
Th1
Severe
P(S1?1)
P(?1)
Th2
P(S2?2)
P(?2)
Moderate
ThN
P(SN?N)
P(?N)
Probability of surviving ingress part of mission
The shorter the dwell, or the longer the revisit
time, the longer will be the reaction time, and
the lower will be the probability of surviving a
threat when it is encountered.
EGRESS is similar
21INGRESS (Success of hunt)
22INGRESS, Cont. (Success of hunt)
Conditional mathematical expectation of total
value of destruction, given that hunter survives
23From Fixed to Dynamic Scan Scheduling
- There are many reasons for having dynamic
scheduling - New targets have been detected and need to be
tracked - The plane (with receivers on board) is moving
and thus the relative illumination times - of various targets have changed (?)
- Terrain masking can suddenly disappear, as the
aircraft travels, thereby exposing aircraft - to being detected
- Schedulers FOM (Figure of Merit) function
changes since some emitters shifted their - operational mode
- Surveillance mode PRF (1 kHz)
- Pulse Doppler PRF (10 to 20 kHz)
- Precision tracking (pencil antenna beam)
pulse-Doppler PRF - Additionally, emitters can change their
characteristics - Changing PRI (staggering), but fixed frequency
- Modulation
- Initial goal detection? tracking varying
emitter parameters? - (As the emitter changes mode (and parameters),
lethality can change Emitter in surveillance
mode detects aircraft vehicle and then - changes mode to precision track. When emitter
reaches a fire-control solution, a surface-to-air
missile is fired, and its radar seeker - begins transmitting.)
- Can simulate varying parameters with a dynamic
process?
24Dynamic Scan Scheduler
25Dynamic Scheduling through Negotiation
- Goal - mission maximize probability of success
of mission - ingress maximize probability of destruction of
target(s) - egress maximize probability of survival (return
home intact) - Goal must translate to a negotiation problem
(conflicting - objectives for negotiating parties
- Sensor agent maximize accuracy of tracking
according to - priorities - initial
goal - maximize probability
of survival - next goal - minimize ratio of
dwell time to revisit time - Threat agent minimize accuracyof tracking -
initial goal - maximize probability
of intercept - next goal - maximize ratio of
dwell time to revisit time
26Negotiation Configuration
. . .
Sensor Agent
Threat Agents
27Schedule Negotiation
- Example (ingress)current threat situation
threat prob. vector - Maximize
- Seen as a resource sharing problem
- resource sensor utilization
- how much of it is available 1
- to be shared between N threat agents
- each of these agents gets a fraction (?n/Tn) of
the resource,where ?n dwell-time allocated and
Tnrevisit time allocated to the agent.
(P(?1), P(?2),P(?N))
Constraint
28Schedule Negotiation (detection)
- Sensor agent minimize (?n/Tn)
- weigh according to threat priority
- as a first approximation
- subject to constraints
- Threat agent maximize (?n/Tn)
- Open issues
- ensure that real-time constraints are met
- how to take probability of threats (I.e P(e))
into account - Dynamic scheduling
- renegotiate the schedule when threat probability
vector changes