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

杨利霞

RouteNet: Leveraging Graph Neural Networks for

! M+ ?  Y) t6 o+ _

Network Modeling and Optimization in SDN

% e/ p1 ^) |7 A$ o: k& ~# `, s- M6 ?
7 p4 |! D! A5 c! M' iNetwork modeling is a key enabler to achieve
effificient network operation in future self-driving Software
' \9 j3 l/ U* J4 ~5 jDefifined Networks. However, we still lack functional network
; Q3 ?) P7 R. K- g% ]+ U' L+ p! i  ]models able to produce accurate predictions of Key Performance
Indicators (KPI) such as delay, jitter or loss at limited cost.
In this paper we propose RouteNet, a novel network model based
) Z8 D! ?& {4 ?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
0 k1 B" f( a+ O1 K: Cability of GNNs to learn and model graph-structured information
1 D0 R* C; \2 z# fand as a result, our model is able to generalize over arbitrary
topologies, routing schemes and traffific intensity. In our eval
7 D/ S' B, l* Fuation, we show that RouteNet is able to predict accurately
/ ?0 Y$ Z) ?2 \9 N0 ~the delay distribution (mean delay and jitter) and loss even in
topologies, routing and traffific unseen in the training (worst case
4 ~0 {" N- `7 U" I/ s7 BMRE = 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.