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

2 d7 n( r& H) }& D
- R* t+ c  H# G# B( X- ?2 o
# V& w7 w$ Q2 XNetwork modeling is a key enabler to achieve
effificient network operation in future self-driving Software
Defifined Networks. However, we still lack functional network
7 x) d% {2 x4 x4 C' C1 Z; \models able to produce accurate predictions of Key Performance
$ Y; H: _$ `$ o+ b4 f' z2 M  ~+ lIndicators (KPI) such as delay, jitter or loss at limited cost.
In this paper we propose RouteNet, a novel network model based
3 D" |8 N5 H! `7 ]on Graph Neural Network (GNN) that is able to understand
+ U4 x9 ~$ e& Athe 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
0 L  w& X* O5 U; r) l0 Kability of GNNs to learn and model graph-structured information
and as a result, our model is able to generalize over arbitrary
topologies, routing schemes and traffific intensity. In our eval
0 F1 s. w4 x' T' Z* Yuation, we show that RouteNet is able to predict accurately
the delay distribution (mean delay and jitter) and loss even in
  b% T; V1 N% d8 T# |( ttopologies, routing and traffific unseen in the training (worst case
) ?4 g" O% L' Y) u1 QMRE = 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.

) O3 H2 K& P& y. b: V, i