TD-STGT: A Spatio-Temporal Graph Transformer for Mobile Traffic Demand Forecasting

Mohamad Alkadamani; Halim Yanikomeroglu

eess.SYSep 6, 2026

TD-STGT: A Spatio-Temporal Graph Transformer for Mobile Traffic Demand Forecasting

Authors: Mohamad Alkadamani, Halim Yanikomeroglu

Organizations: Communications Research Centre Canada, Ottawa, ON, Canada · Department of Systems and Computer Engineering, Carleton University, Ottawa, ON, Canada

Abstract

Fine-grained mobile traffic demand forecasting is essential for long-term planning of 5G and future 6G networks, including radio upgrades, site densification, backhaul expansion, and spectrum activation. This paper proposes the Traffic Demand Spatio-Temporal Graph Transformer (TD-STGT), a graph neural forecasting framework for predicting changes in wireless mobile traffic demand across fine geographic grids. The framework uses a population-scaled demand proxy developed from crowdsourced mobile measurements and daytime population information. Experiments across five Canadian metropolitan regions show that TD-STGT achieves the best performance in forecasting grid-level demand changes, reaching a $ΔR^2$ of 0.462 and reducing $Δ$ RMSE by 5.7% relative to the strongest baseline. The proposed model provides a practical tool for identifying areas with increasing demand pressure and prioritizing future mobile-network capacity upgrades.

Figures & tables

Fig. 1: Temporal evolution of the proposed population-scaled peak-demand proxy.

Fig. 2: Spatial rate of change of the proposed proxy in Toronto.

Fig. 3: Overview of the TD-STGT architecture.

Parameter	Value
Grid nodes $N$	38,246
Input features $F$	18 proxy-derived and auxiliary features
Input window $L$	48 snapshots
Forecast horizon $H$	1 snapshot
Graph neighbors $k$	8
Embedding dimension $d$	128

TABLE I: TD-STGT model configuration

Model	Hyperparameter	Value
Linear ARX	Regularization	Ridge, $\alpha=1.0$
RF	Estimators, depth	300, 18
XGBoost	Estimators, LR, depth	500, 0.03, 6
LSTM	Layers, hidden, dropout	2, 96, 0.2
GRU	Layers, hidden, dropout	2, 96, 0.2
CNN-LSTM	Kernel, hidden, dropout	3, 96, 0.2

TABLE II: Key hyperparameters of baseline models

Proxy variant	Median $\rho$	OLS $R^{2}$	$p$ -value
Observed all-day bytes/hour	0.64	0.60	$<0.001$
Population-scaled peak-demand proxy	0.71	0.66	$<0.001$

TABLE III: Proxy validation against deployed bandwidth

Model	$\Delta R^{2}\uparrow$	$\Delta$ RMSE $\downarrow$	$\Delta$ MAE $\downarrow$
No-change baseline	0.000	0.069	0.047
Linear ARX	0.112	0.065	0.045
RF	0.246	0.060	0.041
XGBoost	0.312	0.057	0.039
LSTM	0.274	0.059	0.040
GRU	0.291	0.058	0.039

TABLE IV: One-step demand-change forecasting performance on the 2025 test set

Fig. 4: Absolute delta-error distributions across models.

Fig. 5: Change-forecasting robustness across 20 random seeds.

Hyperparameter	Setting	$\Delta$ RMSE	$\Delta$ MAE
Learning rate	$5\times 10^{-4}$	0.053	0.036
	$1\times 10^{-3}$	0.050	0.034
	$2\times 10^{-3}$	0.054	0.037
Attention heads	2	0.054	0.037
	4	0.050	0.034
	8	0.052	0.035

TABLE V: Sensitivity analysis for key hyperparameters

Variant	$\Delta R^{2}\uparrow$	$\Delta$ RMSE $\downarrow$	$\Delta$ MAE $\downarrow$
Full TD-STGT	0.462	0.050	0.034
Without spatial module	0.311	0.063	0.043
Without local GAT branch	0.421	0.052	0.036
Without spatial transformer	0.394	0.054	0.037
Without TCN	0.372	0.055	0.038
Without temporal transformer	0.386	0.054	0.037

TABLE VI: Ablation analysis of TD-STGT

Model	Params	Train Time	Inference	$\Delta$ RMSE
	(M)	(min)	(ms/batch)
LSTM	0.31	38	1.7	0.059
GRU	0.27	36	1.5	0.058
CNN-LSTM	0.58	47	2.2	0.056
STGCN	1.12	59	2.9	0.053
TD-STGT	2.86	88	4.4	0.050

TABLE VII: Model complexity and computational cost

Traffic Forecasting

Traffic forecasting predicts future flows, speeds, demand, travel times, and congestion from spatial, temporal, and contextual data.

See all 87 papers

Spatial PDE-aware Selective State-space with Nested Memory for Mobile Traffic Grid Forecasting
Mar 12, 2026Zineddine Bettouche, Khalid Ali, Andreas Fischer +1Spatiotemporal Forecasting Traffic Forecasting
Deggendorf Institute of Technology, Dieter-Görlitz-Platz 1, 94469 Deggendorf
MoDE-Boost: Boosting Shared Mobility Demand with Edge-Ready Prediction Models
Feb 18, 2026Antonios Tziorvas, George S. Theodoropoulos, Yannis TheodoridisSpatiotemporal Forecasting Intelligent Transportation Systems
Department of Informatics, University of Piraeus, Piraeus, Greece.
Spatiotemporal Graph Transformer for Traffic Intelligence in Edge Computing
Aug 4, 2026Laha Ale, Letian Lin, Na Cao +2Edge Computing Traffic Forecasting
School of Computing and Artificial Intelligence, Southwest Jiaotong University, Chengdu, China · SWJTU-Leeds Joint School, Southwest Jiaotong University, Chengdu, China · Key Lab of Information Coding and Transmission, Southwest Jiaotong University, Chengdu, China +1

Mar 12, 2026J/K move · Enter open · S save

Spatial PDE-aware Selective State-space with Nested Memory for Mobile Traffic Grid Forecasting

Zineddine Bettouche, Khalid Ali, Andreas Fischer, Andreas Kassler

Deggendorf Institute of Technology, Dieter-Görlitz-Platz 1, 94469 Deggendorf

Spatiotemporal Forecasting Traffic Forecasting Selective SSMs

Open paper AskPDF arXiv

Traffic forecasting in cellular networks is a challenging spatiotemporal prediction problem due to strong temporal dependencies, spatial heterogeneity across cells, and the need for scalability to large network deployments. Traditional cell-specific models incur prohibitive training and maintenance costs, while global models often fail to capture heterogeneous spatial dynamics. Recent spatiotemporal architectures based on attention or graph neural networks improve accuracy but introduce high computational overhead, limiting their applicability in large-scale or real-time settings. We study spatiotemporal grid forecasting, where each time step is a 2D lattice of traffic values, and predict the next grid patch using previous patches. We propose NeST-S6, a convolutional selective state-space model (SSM) with a spatial PDE-aware core, implemented in a nested learning paradigm: convolutional local spatial mixing feeds a spatial PDE-aware SSM core, while a nested-learning long-term memory is updated by a learned optimizer when one-step prediction errors indicate unmodeled dynamics. On the mobile-traffic grid (Milan dataset) at three resolutions (202, 502, 1002), NeST-S6 attains lower errors than a strong Mamba-family baseline in both single-step and 6-step autoregressive rollouts. Under drift stress tests, our model's nested memory lowers MAE by 48-65% over a no-memory ablation. NeST-S6 also speeds full-grid reconstruction by 32 times and reduces MACs by 4.3 times compared to competitive per-pixel scanning models, while achieving 61% lower per-pixel RMSE.

Abstract

Traffic Forecasting

Explore similar work

Spatial PDE-aware Selective State-space with Nested Memory for Mobile Traffic Grid Forecasting

MoDE-Boost: Boosting Shared Mobility Demand with Edge-Ready Prediction Models

Spatiotemporal Graph Transformer for Traffic Intelligence in Edge Computing