By reference to the derivation of the RLS algorithm in Chapter 3, derive the EWP
algorithm.
Implement the simple scalar FIS algorithm described briefly in the text of section
Then apply this to the Walgett data and check the result using the following
Matlab m-file script, which exploits the dlr routine in CAPTAIN, where z is the
Walgett data vector:
%Forward path filtering
[fit,fitse,par,parse,comp,e,y0]=dlr(z,ones(size(z)),0,qa,...
[],570,100,0);
zr=flipud(par);%reverse order of par filtered estimate
%Simple backward path smoothing of z
[fitb,fitseb,parb,parseb,compb,eb,y0b]=dlr(zr,ones(size(zz)),...
0,qa,[],pard(end),parsed(end)ˆ2,0);
%Optimal FIS for comparison
[fitd,fitsed,pards,parsed,compd,ed,y0d]=dlr(z,ones(size(z)),...
0,qa,[],570,100,1);