RouteNet: Leveraging Graph Neural Networks for Network Modeling and Optimizat...

杨利霞

RouteNet: Leveraging Graph Neural Networks for

Network Modeling and Optimization in SDN

3 C' E2 W8 E! v# {: Q3 Z6 sNetwork modeling is a key enabler to achieve
$ h, g- U* e2 \+ x" D4 a6 m# Geffificient network operation in future self-driving Software
9 D; o7 ?: ~) A  C7 v$ rDefifined Networks. However, we still lack functional network
models able to produce accurate predictions of Key Performance
$ L6 D8 O6 d* ]! r4 f& iIndicators (KPI) such as delay, jitter or loss at limited cost.
9 P. I/ B# ?; k  ~- l) P! Q; l" t% OIn this paper we propose RouteNet, a novel network model based
on Graph Neural Network (GNN) that is able to understand
the complex relationship between topology, routing, and input
traffific to produce accurate estimates of the per-source/destination
per-packet delay distribution and loss. RouteNet leverages the
ability of GNNs to learn and model graph-structured information
and as a result, our model is able to generalize over arbitrary
+ I2 Y* E) |, F) u  A; [' ~+ Z+ \topologies, routing schemes and traffific intensity. In our eval
uation, we show that RouteNet is able to predict accurately
5 l4 [/ o( B) b6 r" _: m( Cthe delay distribution (mean delay and jitter) and loss even in
topologies, routing and traffific unseen in the training (worst case
8 f. F8 [$ d. {8 _MRE = 15.4%). Also, we present several use cases where we
, x: c; ?+ a3 f* ^1 {4 E4 d/ Oleverage the KPI predictions of our GNN model to achieve
# {9 P2 S. x- L* i; neffificient routing optimization and network planning.
- j) Y" g6 f# X# V
: N7 V: ^0 e. F. Z" g
# m: z" s# h. _( o4 h$ F, K