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

" d9 I$ @% N3 e2 ^  B7 J) }
* m, [( w  G" {7 D0 ^2 S/ {/ oNetwork modeling is a key enabler to achieve
: ^) T. U6 d3 A- M9 }; N" Deffificient network operation in future self-driving Software
Defifined Networks. However, we still lack functional network
0 s& x' e4 ?' j. [- y& rmodels able to produce accurate predictions of Key Performance
) l: ?0 a; w3 Y9 o0 QIndicators (KPI) such as delay, jitter or loss at limited cost.
In this paper we propose RouteNet, a novel network model based
) t7 y8 ?5 C6 ^# Y$ M- \on Graph Neural Network (GNN) that is able to understand
the complex relationship between topology, routing, and input
0 q6 Y- h: u) q. _; Y7 y$ T; X! N7 ]traffific to produce accurate estimates of the per-source/destination
per-packet delay distribution and loss. RouteNet leverages the
1 c1 @$ W( l9 l+ m' Vability of GNNs to learn and model graph-structured information
$ a+ p. J4 s) F( ]; kand as a result, our model is able to generalize over arbitrary
topologies, routing schemes and traffific intensity. In our eval
2 o3 H- k: Y4 l3 P; Q9 |uation, we show that RouteNet is able to predict accurately
2 ?/ q2 K- D0 b' {2 t9 L& Z4 J* ~the delay distribution (mean delay and jitter) and loss even in
topologies, routing and traffific unseen in the training (worst case
MRE = 15.4%). Also, we present several use cases where we
leverage the KPI predictions of our GNN model to achieve
effificient routing optimization and network planning.
7 q, y9 ^0 {3 g- a
0 W+ R7 }9 r3 {/ F' U# x- F+ a