Sensiverse: A dataset for ISAC study

Jiajin Luo,Baojian Zhou,Yang Yu,Ping Zhang,Xiaohui Peng,Jianglei Ma,Peiying Zhu,Jianmin Lu,Wen Tong

2023-08-26

Abstract:In order to address the lack of applicable channel models for ISAC research and evaluation, we release Sensiverse, a dataset that can be used for ISAC research. In this paper, we present the method of generating Sensiverse, including the acquisition and formatting of the 3D scene models, the generation of the channel data and associations with Tx/Rx deployment. The file structure and usage of the dataset are also described, and finally the use of the dataset is illustrated with examples through the evaluation of use cases such as 3D environment reconstruction and moving targets.

Signal Processing

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the lack of applicable channel models suitable for integrated sensing and communication (ISAC) research and evaluation. Specifically, the existing communication channel models cannot be directly used to evaluate sensing performance, because the requirements of sensing services (such as measurement accuracy and resolution) are different from those of communication services. In addition, traditional stochastic channel models are insufficient in simulating real - physical environments, especially in imaging scenarios, and may not accurately reflect characteristics such as secondary echoes in the environment.

To address these problems, the authors constructed a sensing channel data set named Sensiverse. This data set includes channel data in different scenarios and provides the following improvements:

1. **Support for multi - scenario and multi - mode sensing**: Sensiverse supports moving target detection, 3D environmental imaging and reconstruction, and the evaluation of integrated sensing and communication solutions.
2. **Improvement of the accuracy of evaluation results**: By providing a large number of channel samples, the problem of evaluation results not converging to the real performance due to a limited number of samples is avoided.
3. **Reduction of computational complexity**: The data set is generated and stored offline, trading storage for computational resources and reducing the consumption of computational resources required for subsequent large - scale simulation evaluations.
4. **Support for calibration and training of hybrid channel models**: It can be used for calibration and training of hybrid channel models, as well as training and testing of machine learning in the sensing field.

The paper details the generation method, file structure, and usage examples of Sensiverse, aiming to provide a powerful tool for future ISAC research and evaluation.

### Main problem summary:
- **Limitations of existing communication channel models**: They cannot be directly used for sensing performance evaluation.
- **Deficiencies of traditional stochastic channel models**: They cannot accurately reflect the real - physical environment, especially in imaging scenarios.
- **Immature methods for evaluating ISAC system performance**: A more comprehensive data set is required to support the evaluation.

By constructing the Sensiverse data set, the authors hope to fill this gap and provide strong support for ISAC research and applications.