Non-Collaborative Distributed Network Optimization
|
Leveraging our ground-breaking research into
game theoretic analysis and design of cognitive
radio net-works, CRT has developed a suite of
distributed non-collaborative algorithms which
converge to stable radio resource allocations (e.g.,
power, frequency, modulation, beam patterns)
which optimize network performance without
requiring a centralized controller or coordination
between distributed controllers.
|
|
Highly Scalable
Because our algorithms work with instead of
against network interactions, there is no need to
- coordinate adaptations between controllers
- distribute observations between controllers
- message a centralized controller
- distribute clocks
|
All this adds up to networks whose complexity is
in-dependent of scale (our basic ad-hoc suite) or
scales linearly with network density (our traffic
re-active and beamforming suites).
|
For our algorithms, selfishness is
socially optimal
|
Widely Applicable
Because our algorithms encompass all traditional PHY resources, e.g., channel,
power, modulation, transmit/receive beam patterns, and have been designed to
accommodate a wide variety of scenarios, e.g., MIMO systems, subcarrier allocation,
our algorithms can optimize networks running a wide variety of waveforms
including:802.11a/b/g/n, Mobile/Fixed WiMAX, 802.22, LTE, UMB, TD-SCDMA, and
Zigbee. This is enhanced by our commitment to portable software design.
|
Dramatic Improvements in Network Performance
Achieve pre-planned like performance, but optimized in real-time to changing network
conditions.
|
|
