The increasing numbers of devices, functions, traffic types and volumes can create headaches for network operators.
And the need to be flexible and remain competitive raises another set of equally imposing challenges. New network protocols must be evaluated against a larger set of criteria – including energy expenditures, localization and multi-homing. Care must also be taken that the very solutions to address these issues will not introduce self-destructive levels of complexity.
A team of Nokia Bell Labs researchers has made significant progress in automating performance optimization and adaptive operation by building on existing concepts in self-organization, stochastic gradient search and multi-armed bandits theory.
Modeling the dynamics of LTE HetNets, researchers have devised ways to optimize hand-off protocols and methods for interference management. This research has yielded optimal bias and advanced scheduling schemes in Enhanced Inter-Cell Interference Coordination (eICIC) as well as in other Layer 2 approaches (Carrier Aggregation with Soft Frequency Reuse and Coordinated Scheduling / Coordinated Beamforming).
Gordon Wilfong has had success in modeling the dynamic behavior of Border Gateway Protocols (BGP) as a static graph game – identifying a fundamental structure, the Dispute Wheel, whose absence guarantees convergence. This property of BGP has been widely adopted.
Anwar Walid has developed a theory for energy efficient congestion control. It allows for the design of efficient sleep-mode mechanisms that co-exist with traditional single path TCP and the emerging multi-path TCP.