The distributed adaptive signal fusion (DASF) algorithm framework implementation based on [1] and [2].
Given an optimization problem fitting the DASF framework:
P: min_X f_hat ( X'*y(t), X'*B, X'*Gamma*X ) = f(X)
s.t. h_j ( X'*y(t), X'*B, X'*Gamma*X ) <= 0 for inequalities j,
h_j ( X'*y(t), X'*B, X'*Gamma*X ) = 0 for equalities j,
the DASF algorithm solves the problem in a distributed setting such as a wireless sensor network consisting of nodes connected to each other in a certain way. This is done by creating a local problem at node q and iteration i and has the advantage that the local problem is a parameterized version of problem P. Therefore, a solver for problem P is used for the distributed implementation.
Note: There can be more than one y(t), B and Gamma which are not represented for conciseness.
The dasf function implements the DASF algorithm and is called in the following way:
Matlab:
[X_est,norm_diff,norm_err,f_seq]=dasf(prob_params,data,...
@prob_solver,conv,@prob_select_sol,@prob_eval)
Python:
X_est,norm_diff,norm_err,f_seq=dasf(prob_params,data,...
prob_solver,conv,prob_select_sol,prob_eval)
References:
[1] C. A. Musluoglu and A. Bertrand "A Unified Algorithmic Framework for Distributed Adaptive Signal and Feature Fusion Problems - Part I: Algorithm Derivation", IEEE Transactions on Signal Processing, 2023, doi: https://doi.org/10.1109/TSP.2023.3275272.
[2] C. A. Musluoglu, C. Hovine and A. Bertrand "A Unified Algorithmic Framework for Distributed Adaptive Signal and Feature Fusion Problems - Part II: Convergence Properties", IEEE Transactions on Signal Processing, 2023, doi: https://doi.org/10.1109/TSP.2023.3275273.