Jin, DanqiChen, JieRichard, CedricChen, JingdongSayed, Ali H.2020-12-242020-12-242020-12-242020-01-0110.1109/TSIPN.2020.3038017https://infoscience.epfl.ch/handle/20.500.14299/174317WOS:000597140600002Combining diffusion strategies with complementary properties enables enhanced performance when they can be run simultaneously. In this article, we first propose two schemes for the convex combination of two diffusion strategies, namely, the power-normalized scheme and the sign-regressor scheme. Then, we conduct theoretical analysis for one of the schemes, i.e., the power-normalized one. An analysis of universality shows that it cannot perform worse than any of its component strategies in terms of the excess mean-square-error (EMSE) at steady state, and sometimes even better. An analysis of stability also reveals that it is more stable than affine combination schemes already proposed by the authors in the literature. Next, several adjustments are proposed to further improve the performance of convex combination schemes. A discussion about the computational and communication complexity is provided, as well as a comparison between convex and affine combination schemes. Finally, simulation results are shown to demonstrate their effectiveness, the accuracy of the theoretical results, and the improved stability of the convex power-normalized scheme over the affine one.Engineering, Electrical & ElectronicTelecommunicationsEngineeringadaptive fusion strategyconvex combinationdiffusion strategydistributed optimizationperformance analysisleast-mean squaresaffine combinationadaptive filterslmsperformanceadaptationtransientsparsitytext::journal::journal article::research article