Dual-Stage LOS/NLOS Classification Across Changing UWB Environments
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing LOS/NLOS classification models in Ultra Wideband (UWB) systems are not robust across different environments, leading to performance degradation when applied in new environments, and require retraining for each environment change.
Innovation Solution
A two-stage classification system is proposed, comprising a pre-classification unit to determine the environment class and a subsequent LOS/NLOS classification model, where the environment class is used as a modulation parameter to enhance model performance across varying environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a specialized LOS/NLOS classification model is trained for a specific environment, then the classification accuracy is improved in that environment, but the model performance degrades when applied to other environments
Solution Approach 1:
The classification system is segmented into two independent stages: an environment classification model that identifies the current environment type, and a LOS/NLOS classification model that performs the actual classification. This segmentation allows each model to specialize in its specific task, with the environment model adapting to different settings while the LOS/NLOS model maintains consistent classification capability across all environments.
Solution Approach 2:
The environment classification model acts as an intermediary between the input channel impulse response and the LOS/NLOS classification model. It first characterizes the environment and passes this information to the LOS/NLOS classifier, enabling the system to adapt to different environments without retraining the LOS/NLOS classification logic.
2Adaptability or versatility
If a multi-environment classification model is attempted, then the adaptability is improved, but the performance turns out to be poor
Solution Approach 1:
Rather than creating a single multi-environment model, the system segments the classification task into environment identification and LOS/NLOS determination. This allows the system to achieve multi-environment adaptability while maintaining high performance in each specific environment through the specialized LOS/NLOS classifier.
Solution Approach 2:
The system dynamically adapts to different environments by first classifying the environment type and then using that information to guide the LOS/NLOS classification. This dynamic approach allows the system to maintain optimal performance across varying environments without requiring a single static multi-environment model.
3Loss of time
If transfer learning is used to update models in new environments, then the learning time is reduced, but new learning and model updates are still needed for each distinct environment
Solution Approach 1:
The system segments the adaptation process so that only the environment classification model needs to be updated when transitioning to a new environment, while the LOS/NLOS classification model remains unchanged. This reduces both the time and complexity of model updates compared to retraining entire multi-environment models.
Solution Approach 2:
The environment classification model serves as an adaptable intermediary that can be quickly updated for new environments through transfer learning, while the core LOS/NLOS classification logic remains stable and requires no retraining, thereby reducing overall system complexity and update requirements.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
A multi-environment LOS/NLOS classifier 24 is proposed in a ULB telemetry device. It has two stages: an environment classifier (241) and a LOS/NLOS classifier (242). The environment classifier includes a CNN (241a) fed by the CIR of a received radio signal, cascaded with an MLP (241b) also fed by a set (SET_1) of statistical features extracted from the CIR by a dedicated unit (23). The environment classifier provides an environment class (ENV_ID) to the LOS/NLOS classifier, typically an LP, which also receives as input the output of the CNN and another set (SET_2) of statistical and physical features extracted from the CIR. The environment class (ENV_ID) can form a model input of the MLP operating as a LOS/NLOS classifier, or be a modulation parameter of the MLP, typically a bias or a weight modification factor of the MLP.