Game Theoretical Framework based algorithm is a hotcake nowadays in wireless communication research fields owing to its multiple application. Gaming algorithm is impressive in preparing economic model but there might be some obstacles in implementing the algorithm practically. I want to know about those limitations which may affect implementation of this algorithm practically in a wireless network.
First, the communications overhead from a distributed algorithm that needs several iterations to converge to a solution, typically makes it impractical for a direct application of the algorithm.
Second, a distributed algorithm with built-in fairness considerations will typically be vulnerable to malicious nodes that misbehave.
Third, the interactions with transport layer protocols such as TCP (which have different fairness mechanisms) will make it difficult to obtain a balanced solution for most situations.
As Fausto alludes to, the overhead of maintaining the data needed to enforce the bandwidth commitments or policies become the problem. It becomes especially difficult if you need some shared state to be achieved.
So you need to pick your battles.
People tend to be overly clever in trying to design these systems. They want 100% accurate, 100% timely accounting information or perfect enforcement of flow control and jitter. The inflexibility of perfection creates constraints on design and behavior, some of which tend to be disastrous when the system is under attack.
Often its beneficial to turn the constraints inside out.
- Badges
- Science method