A New PCA Based High Dimensional Reinforcement Learning Multi Agent Approach For Traffic Shaping In Routers
In this paper, the concept of Principle Component Analysis (PCA) is invoked besides reinforcement learning and
multi-agent systems to develop a novel intelligent high dimensional reinforcement learning traffic shaper for dynamic and
real time allocation of the rate of generation of tokens in a Token Bucket algorithm instead of static allocation of this
parameter. This implementation when is compared to our previous work where a simple reinforcement learning traffic
shaper was developed, the better and more reasonable utilization of bandwidth and less traffic overload in other parts of the
network is more appeared. Indeed, the imposed PCA on the inputs of the reinforcement algorithm gives this ability to the
traffic shaping agents to use more vital parameters of the network in their decision process without any concern about
exceeding the volume of the calculations and the time. This novel work is also valuable in this aspect that it offers a high
dimensional functionality to the reinforcement learning algorithm in the context of multi-agent systems where incrementing
dimension is a practical limitation. These methods are implemented in our previous proposed intelligent simulation
environment to be able to compare better the performance metrics. The results obtained from this simulation environment
show satisfactory behaviors from the aspects of keeping whole dropping probability low while injecting as many packets as
possible into the network in order to utilize the available bandwidth as much as possible.
Keywords- High Dimensional, Principle Component Analysis (PCA), Traffic Shaping In Routers.