Bregman-Based Inexact Excessive Gap Method for Multiservice

Bregman-Based Inexact Excessive Gap Method for Multiservice Resource
In order to meet the explosive increasing demand of user application data
in modern wireless networks, a variety of multiservice resource allocation
algorithms have been proposed in the literature. Most of them can be
modeled as optimization problems of minimizing a summation function
indicated as Σi=1Nfi(xi) with additive nonlinear coupling inequality
constraints. The existing subgradient methods can only achieve a
convergence rate of O(1/√k), which is quite slow for handling big user data
generated from modern heterogeneous wireless networks. To develop
more efficient multiservice resource allocation algorithms, we consider the
regularized Lagrangian function with smoothing accelerated techniques.
Specifically, in this paper, we extend the previous research that mainly
focuses on linear coupling equality constraints to a challenging scenario
with nonlinear coupling inequality constraints. To solve the problem, we
propose and analyze a Bregman-based inexact excessive gap (BIEG)
algorithm, which, by rigorous mathematical proofs, can asymptotically
achieve a faster convergence rate of O(1/k). Furthermore, the BIEG method
is applied to develop a novel multiservice resource allocation algorithm,
namely, BIEG-RA, which combines the accuracy control mechanism with
the Bregman projection technique. Numerical results verify its fast
convergence rate in heterogeneous wireless networks.