// BCJR.h -- Compute the MAP bit decisions using the BCJR algorithm// Todd K. Moon// Feb 17, 2003// Copyright 2004 by Todd K. Moon// Permission is granted to use this program/data// for educational/research only#ifndef BCJR_H#define BCJR_h#include "BinConv.h"#include "matalloc.h"class BCJR {protected:   double **alpha;				// [L+1][Q] forward probabilities   double **beta;				// [L+1][Q] backward probabilities   double **Pp;					// [L][numbranch] posterior bit probability   int nu;						// constraint length   int Q;						// Q = 2^nu; number of states   int L;						// total number of trellis stages   int n;						// dimension of received signal vector   double sigma2;				// noise variance in each component   int N1trans;					// number of state transitions where x=1   unsigned int **S;			// (p,q): S[1][p]=q or S[0][p] = q   unsigned int **prevstate;    // [Q][Q] list of previous states   unsigned int **inputfrom;	// [Q][numbranches] -- inputs   double ***outputmat;			// [Q][numbranches][n] -- outputs   int k;						// number of inputs   int numbranches;				// number of branches = 2^numinputs   void buildprev(BinConv &encoder); // build information about trellis   void buildoutputmat(BinConv & encoder);   double gammakpq(int k, int p, int q, int inp, double Ppq, const double *r);   // Compute the transition probability;   // Assumes that (p,q) is a valid   // state transition.   // inp is the input at state p,    // Ppq is the probability of the input that causes the transition p to q   double gammakpq2(int k, int p, int q, int i,const double *r);   // This version is used for computing the "extrinsic probabilities."   // It uses only the "nonsystematic" part of r   // and does not use the prior probabilities.public:   BCJR(BinConv  &encoder, int in_L,double in_sigma2);   ~BCJR() {FREEMATRIX(S); FREEMATRIX(inputfrom); FREETENSOR(outputmat,Q);   FREEMATRIX(alpha);  FREEMATRIX(beta);  delete[] Pp;   };   void alphabeta(const double **r,const double **prior, int finalstate=0);   // compute the forward and backward passes   // r[i][n] -- received sequence   // prior[numbranch] -- the prior probabilities at time k for each input   // e.g. prior[0] = P(u_k=0), prior[1] = P(u_k=1)   double **MAP1(const double **r, const double **prior, int finalstate=0);   // Compute the LLR using the entire r vector (not systematic)   // r[i][n] -- received sequence   // set finalstate<0 for    // uniform distribution, or set finalstate to actual state   // returns a pointer to the posterior probability   double **MAP2(const double **r, const double **prior, int finalstate=0);   // Compute the LLR using nonsystematic part and systematic parts separately   // Returns a pointer to the extrinsic part of the probability   // which resides in Pp};#endif/*Local Variables:compile-command: "g++ -c -g BCJR.cc"End:*/