DC-STGCN: Dual-Channel Based Graph Convolutional Networks for Network Traffic Forecasting

Abstract

Network traffic forecasting is essential for efficient network management and planning. Accurate long-term forecasting models are also essential for proactive control of upcoming congestion events. Due to the complex spatial-temporal dependencies between traffic flows, traditional time series forecasting models are often unable to fully extract the spatial-temporal characteristics between the traffic flows. To address this issue, we propose a novel dual-channel based graph convolutional network (DC-STGCN) model. The proposed model consists of two temporal components that characterize the daily and weekly correlation of the network traffic. Each of these two components contains a spatial-temporal characteristics extraction module consisting of a dual-channel graph convolutional network (DCGCN) and a gated recurrent unit (GRU). The DCGCN further consists of an adjacency feature extraction module (AGCN) and a correlation feature extraction module (PGCN) to capture the connectivity between nodes and the proximity correlation, respectively. The GRU further extracts the temporal characteristics of the traffic. The experimental results based on real network data sets show that the prediction accuracy of the DC-STGCN model overperforms the existing baseline and is capable of making long-term predictions.

Prediction model for short‐term traffic flow based on a K‐means‐gated recurrent unit combination

“…Pan et al. [52] constructed a dual‐channel deep learning framework, which consists of dual‐channel GCN and GRU. This framework can characterize the temporal correlation of traffic flow data.…”

Graph-based Deep Learning for Communication Networks: A Survey

“…Computer Networks [13,14] Electronics [15] IEEE Access [16,17] IEEE Communications Letters [18,19,20,21] IEEE Internet of Things Journal [22] IEEE Journal on Selected Areas in Communications [23,24,25,26] IEEE Systems Journal [27] IEEE Transactions on Industrial Informatics [28] IEEE Transactions on Information Forensics and Security [29] IEEE Transactions on Mobile Computing [30,31] IEEE Transactions on Network Science and Engineering [32] IEEE Transactions on Network and Service Management [33] IEEE Transactions on Signal Processing [34] IEEE Transactions on Vehicular Technology [35] IEEE Transactions on Wireless Communications [36,37] International Journal of Network Management [38] Performance Evaluation [39] Sensors [40] Transactions on Emerging Telecommunications Technologies [41] conducting a thorough literature search on the graph-based models. For now, we would give a short introduction for the GNNs used in the surveyed studies.…”

“…GAT [11] (ICLR 2018) GN [12] (arXiv 2018) GE [13] (IEEE Trans Cybern 2019) GIN [14] (ICLR 2019) HetGAT [15] (WWW 2019) We first introduce the Graph Embedding (GE) models. In mathematics, embedding is a mapping function f : X → Y , in which a point in one space X is mapped to another space Y .…”

“…Traffic prediction is also considered in cellular networks, with GNN-based solutions being proposed in recent years [50,31,15,30]. As a prediction problem, the temporal dependencies may be modeled by a recurrent neural network, e.g., Long Short Term Memory (LSTM) or Gated Recurrent Unit (GRU).…”

“…As an improvement over baselines, GNN is capable of modeling the spatial correlation between different nodes, e.g., a cell tower or an access point. Different structures have been explored in existing studies, e.g., GAT in [50,31], GCN in [15], and GraphSAGE in [30].…”