130 likes | 330 Views
Scheduling for Adaptive Weather Sensing Using Phased Array Radar. Tian-You Yu 1,2 , Ricardo Reinoso-Rondinel 1,2 and Sebastian Torres 2,3,4 School of Electrical & Computer Engineering, University of Oklahoma, USA Atmospheric Radar Research Center, University of Oklahoma, USA
E N D
Scheduling for Adaptive Weather Sensing Using Phased Array Radar Tian-You Yu1,2, Ricardo Reinoso-Rondinel1,2 and Sebastian Torres2,3,4 School of Electrical & Computer Engineering, University of Oklahoma, USA Atmospheric Radar Research Center, University of Oklahoma, USA Cooperative Institute for Mesoscale Meteorological Studies, University of Oklahoma, USA NOAA/OAR National Severe Storms Laboratory, USA This work was primarily supported by NOAA/NSSL under cooperative agreement of NA17RJ1227. Part of this work was supported by the DOD EPSCoR grant N00014-06-1-0590. KU-OU Symposium, November 2009
Outlines • Introduction and Motivation • Quality Measures for Adaptive Weather Sensing: Revisit Improvement Factor and Acquisition Time • Time Balance Scheduling • Simulation Results • Summary
Introduction and Motivation KTLX Observations • WSR-88D VCP scan • Tracking and surveillance are performed using the same pattern • The update time of multiple storms is the same • PAR adaptive scan • Tracking and surveillance can be performed using different patterns • The update time of multiple storms can be adaptive to extract maximum information of storms Adaptive Weather Sensing Radar Function: storm tracking and surveillance Radar task: tracking one storm cell or surveillance Goal: tracking multiple storms with fast and independent update time while surveillance is maintained.
Storm Tracking and Surveillance Tasks Tracking task: Task time Occupancy: Update time • Surveillance task: • In order to maximize the usage of radar resource, the surveillance is only executed when radar is idle. • Unlike the tracking task, the surveillance task can be interlaced to allow maximum flexibility.
Quality Measure I: Revisit Improvement Factor Given the same accuracy as WSR-88D
Quality Measure II: Acquisition Time Acquisition Time: The minimum time for each task is executed at least once
Scheduling Multi-task: Time Balance (TB) Step 1. Storm tracking: number of cells (N), task time (Ti), and update time (Ui) of cell i, i= 1,2,…,N. Surveillance: task fragment time (TF) Each tracing task is assigned a time balance variable Step 2. Set time balance (TB) to zero for new tasks Step 3. Any TB(k)0 k= 1,2,…,N no ≥ yes Step 4. Choose task with maximum TB : i = argmax TB(k) k Step 5. Decrement TBof task i by its update time: TB(i)=TB(i)-Ui Step 8. Schedule surveillance fragment Step 6. Schedule storm tracking task i Step 7. Increment TBof all tasks by task time of task i: TB(k)=TB(k) +Ti k= 1,2,…,N Step 9. Increment TB of all tasks by task fragment time TB(k) = TB(k) + TF k = 1,2,…,N
O1= 50% Demo of TB to Schedule Multi-task T2 O2 = 40% T1 U2 U1 s3TF s1TF s2TF Surveillance D1 T2 D1 U1 U2 T2
Simulated Scan Comparison: WSR-88D and PAR • 2008-04-23 KTLX data from 01:18:43 to 03:03:44 UTC PAR WSR-88D 90o sector
Summary • PAR is capable of performing surveillance and tacking of multiple storm cells independently and adaptively. The concept of Time Balance was introduced to schedule these tasking that are competing for radar resources. • Two quality measures were introduced: revisit improvement factor and acquisition time (can be optimized independently based on user’s need). The trade-off between these two measurements were demonstrated by both theory and simulations. • Scheduling multiple tasks for adaptive sensing was demonstrated using interpolated WSR-88D data. • Results suggest that the improvement factor and acquisition time gained by adaptive sensing can be realized by TB scheduling