Parallel multi-layer cognitive network optimization

摘要

System (300) and methods (400, 600) for providing a Cognitive Network (“CN”). The methods involve: partially solving Multi-Objective Optimization Algorithms (“MOOAs”) for Protocol Stack Layers (“PSLs”) using initialization parameters generated based on project requirements (572); and monitoring the convergence behaviors of MOOAs (584) to identify when solutions (106) thereof start to converge toward Pareto-Optimal solutions (104). In response to said identification, a convergence of a solution trajectory for at least one MOOA is “biased” so that compatible non-dominated solutions are generated at PSLs. A Pareto Front (100) for each PSL is determined by generating remaining solutions for MOOAs. The Pareto Fronts are analyzed in aggregate to develop Best Overall Network Solutions (“BONSs”). BONSs are ranked according to a pre-defined criteria. A Top Ranked Solution (“TRS”) is identified for BONSs that complies with current regulatory/project policies. Configuration parameters are computed for PSL protocols that enable implementation of TRS within CN.

主权项

1. A method for providing a cognitive network, comprising:
partially solving, by at least one network node of a plurality of network nodes, a plurality of Multi-Objective Optimization (“MOO”) algorithms for a plurality of protocol stack layers using initialization parameters generated based on project requirements, where the plurality of MOO algorithms are to be collectively used to optimize performance of the cognitive network; monitoring, by at least said network node, a convergence behavior of said MOO algorithms to identify when solutions thereof start to converge toward Pareto-Optimal solutions; biasing, by at least said network node, a convergence of a solution trajectory for at least a first MOO algorithm of said MOO algorithms so that compatible non-dominated solutions are generated at each of said plurality of protocol stack layers; determining, by at least said network node, a Pareto Front for each protocol stack layer of said plurality of protocol stack layers in a protocol stack by generating all remaining solutions for said plurality of MOO algorithms, where (1) said plurality of MOO algorithms are respectively solved using respective ones of said initialization parameters and (2) at least two of the plurality of MOO algorithms are different for at least two of said plurality of protocol stack layers; analyzing, by at least said network node, said Pareto Fronts previously determined for said plurality of protocol stack layers in aggregate to develop a plurality of best overall network solutions; ranking, by at least said network node, said plurality of best overall network solutions according to a pre-defined criteria; identifying, by at least said network node, a top ranked solution for said plurality of best overall network solutions that complies with current regulatory and mission policies; computing, by at least said network node, configuration parameters for protocols of said protocol stack layers that enable implementation of said top ranked solution within said cognitive network; and dynamically re-configuring network resources of said protocol stack layers in accordance with said configuration parameters.