“An Investigation into the Temporal Correlation at the ASF Monitor Sites” by Prof. Peter Swaszek, URI/USCGA Dr. Gregory Johnson, Alion Capt. Richard Hartnett, USCGA Dr. Sherman Lo, Stanford
or “A Partial Answer to David Last’s Question on Monitor Spacing Requirements”
Background ILA-35 (2006) – “Warping Time and Space: Spatial Correlation of Temporal Variations” –Seasonal Monitor Network Sites, equipment, software –Spatial Correlation Several anecdotal examples –ASF Filtering Reduce receiver noise effects
Prior Conclusions There is an obvious correlation in the ASFs of nearby sites –Depends on local topography Land-path stations experience more variation –Most extreme variations occur in winter Placement of monitors for dLoran will be dependent upon worst-case “correlation” –Winter in the NorthEast is the long pole
ILA-36 (today) Look at some of the available data –2 new sites –Some sites collecting over almost 2 years More on temporal correlation including error effects –Statistical measures –Error performance
The Seasonal Monitors circa Oct. 2007
Sites Monitored at CGA USCGA, New London CT URI, Kingston RI Volpe, Cambridge MA FAATC, Atlantic City NJ OU, Athens OH Staten Island, NY Goodspeed (CT) New Haven (CT) NEW !!!
Seasonal Monitor Sites 12/22/ MANY MILES
Shorter Baselines Distances 9
Purposes of Monitor Network Analysis of ASF variation for aviation –Center of range studies –Bounds on error dLoran system component for HEA –ASF updates to LSU –Broadcast out on LDC Sherman’s presentation next Greg’s presentation tomorrow
What’s New Today We have lots more data, some on shorter baselines –Includes pre/post-TOT transition –2 summers/winters for the early sites Examine statistics versus distance Examine position error performance of dLoran versus distance
Some ASF Data
Typical ASF Data
ASF Data from Monitor Sites Have long assumed that the ASF can be decomposed into 3 independent, additive terms: –Spatial term –Temporal term –Directional term for a moving antenna For further visuals, we remove (zero out) the spatial term –Temporal term forced to mean of zero –Directional term assumed to be zero
Typical Temporal Term
Some Comparisons Seasonal differences –Summer (June1 – August 31) –Winter (January 1 – March 31) Two year repeatability “Correlation” site-to-site –High –Low ASF differences
Our Winter/Summer Definition
Repeatability of ASFs – 2 Years at One Site
Repeatability – Zoom of Summer
Repeatability – Zoom into Winter
Site-to-Site, Strong Correlation
Site-to-Site, Weaker Correlation
Differences of the ASFs
and
Statistics What’s relevant to compute? Correlation coefficient is one option –ρ = 1 just means a “linear” relationship Ignores scaling and offset Not relevant for error analysis Will look at average differences in ASF
Measure “spread” of differences in ASF by standard deviation of differences –Tabulate average standard deviation of differences –Focus on pairwise characteristics of close sites – short baselines only
Table of Results (nanosec) Monitor site pair Distance km Yearly average Summer average Winter average CGA/GSPD HVN/GSPD CGA/URI CGA/HVN URI/GSPD URI/TSC ACY/OU
Position Error Performance How far away from a monitor site are the ASFs good enough for dLoran? –Measure above is unclear –Anecdotal evidence from harbor testing Approach – identify position error due to mismatch –Consider one monitor site as a “mobile” receiver –Use ASFs from second site in position solution
Example –URI & TSC ASFs SWAP ASFs
Example SUMMER WINTER
Performance Results Average over time –All year, winter, summer Tabulate 95% error radii Focus on pairwise characteristics of close sites – short baselines – only
Best Site-to-Site Performance SUMMER WINTER
95% Error Radius vs Distance
Conclusions/Future While ASFs are clearly correlated at nearby sites, position performance is sensitive to mismatch –Close spacing seems necessary for HEA –dLoran for aviation could accept wider spacing –Error budget needs to also include receiver noise and spatial ASF components Will continue collecting and testing data – Get shorter baseline data (along coastline) from PIG/LSU sites Point Allerton (MA) Sandy Hook (NJ)