Litcius/Paper detail

Transformer based Radio Map Prediction Model for Dense Urban Environments

Yu Tian, Shuai Yuan, Weisheng Chen, Naijin Liu

20212021 13th International Symposium on Antennas, Propagation and EM Theory (ISAPE)36 citationsDOI

Abstract

Radio map prediction (RMP) is one of the key technologies to improve spectrum efficiency. In this paper, a novel deep learning model termed as RadioTrans is proposed for RMP task. Specifically, Transformer modules are used to capture the long-range spatial relationship in radio wave propagation. Furthermore, a Grid Anchor technique is proposed to better represent the relative position of the radiation source, destination and environment. The effectiveness of proposed method is verified on an urban radio wave propagation dataset. Compared with state-of-the-art deep learning RMP model, RadioTrans improve the prediction accuracy by 27.3%. Compared with the well-known ray-tracing based method, the prediction speed is increased by 4 orders of magnitude. Code is released at [email protected]:OXSLAB/RadioTrans_Official.git.

Topics & Concepts

Computer scienceTransformerRadio propagationGridDeep learningArtificial intelligenceRay tracing (physics)Code (set theory)Radio waveRangingPosition (finance)TelecommunicationsEngineeringElectrical engineeringGeologyEconomicsSet (abstract data type)PhysicsProgramming languageQuantum mechanicsGeodesyVoltageFinanceMillimeter-Wave Propagation and ModelingRadio Wave Propagation StudiesTelecommunications and Broadcasting Technologies
Transformer based Radio Map Prediction Model for Dense Urban Environments | Litcius