Periodic and non-concurrent error detection and identification in one-hot encoded FSMs (Q705464)

The discrete time finite state machine (FSM) with state transition faults is considered. The FSM is described by set state \(Q=\{q_1,\dots,q_N\}\), input state \(X=\{x_1,\dots,x_U\}\) and output state \(Y=\{y_1,\dots,y_J\}\). In an one-hot encoded FSM with \(N\) states the current state is represented by \(N\)-dimensional binary vector \(q[t]=(0,\dots,1,\dots,0)\) which \(j\)-th nonzero entry is equal to '1' if the state is \(q_j\). This representation allows to consider the next-state function of FSM as an equation \[ q[t+1]=A_{x[t]} q[t], \] where \(A_{x[t]}\) is \((N \times N)\)-matrix such that each of its columns has exactly one nonzero entry with value '1'. This representation allows to use algebraic methods. To detect and identify state transition faults an embedding system is constructed with \((N+m)\)-dimensional state vector \(\xi[t]\) which is described by the next-state function \[ \xi[t+1]={\mathcal A}_{\xi[t]} \xi[t], \] so as the original state vector can be found by the relationship \(q[t]=L \xi[t]\) and the faults can be checked by calculating signatures of \(\xi[t]\). Traditionally concurrent checks of faults are used, that are performed at the end of each step \(t\). In the article periodic error detection is considered, i.e. at the end of each \(L\) steps. This approach is more beneficial in case of rare errors. The appropriate constructions of vector \(\xi[t]\) and matrices \({\mathcal A}_{\xi[t]}\) are given in the article.