As a critical component of the 5G-Advanced and 6G technology systems, passive Internet of Things (IoT) primarily operates through technologies such as backscatter communication, radio frequency energy harvesting, and low-power computing. It is seen as a crucial element in realizing the vision of “trillion-scale” interconnection. Device-free localization based on wireless signals serves as a foundational aspect of wireless intelligent sensing technology. With the development and impetus of passive IoT, wireless signal-based device-free indoor localization technology has become an inevitable trend in the advancement of indoor localization techniques. For device-free multi-target localization, the most direct approach involves creating a dedicated fingerprint database for multiple targets. However, this requires traversing various combinations of target arrangements at reference locations, resulting in an exponential increase in the number of targets. This leads to substantial human effort and time consumption during the database construction process. Additionally, during matching localization, there is an increase in computational complexity. To mitigate data and computational complexity while saving time and human resources, a plausible strategy involves using a single-target fingerprint database for multi-target localization. To tackle the fingerprint overlay error interference in multi-target localization using a single-target fingerprint database, this chapter introduces a device-free localization method based on sparse group lasso. During the initial coarse localization phase, the method assesses the matching degree of each target within the reference areas. Subsequently, the fine localization process is formulated as a polynomial optimization problem integrating sparse representation and a regularization term. Notably, the matching degrees serve as weights for the regularization term. This chapter presents methodologies for parameter adjustments and metrics for evaluating localization performance.