This paper studies a WiFi indoor localisation technique based on using a deep learning model and its transfer strategies. We take CSI packets collected via the WiFi standard channel sounding as the training dataset and verify the CNN model on the subsets collected in three experimental environments. We achieve a localisation accuracy of 46.55 cm in an ideal (6.5m × 2.5m) office with no obstacles, 58.30 cm in an office with obstacles, and 102.8 cm in a sports hall (40 × 35m). Then, we evaluate the transfer ability of the proposed model to different environments. The experimental results show that, for a trained localisation model, feature extraction layers can be directly transferred to other models and only the fully connected layers need to be retrained to achieve the same baseline accuracy with non-transferred base models. This can save 60% of the training parameters and reduce the training time by more than half. Finally, an ablation study of the training dataset shows that, in both office and sport hall scenarios, after reusing the feature extraction layers of the base model, only 55% of the training data is required to obtain the models’ accuracy similar to the base models.
Designing a reliable and highly accurate indoor localization system is challenging due to the non-uniformity of indoor spaces, multipath fading, and satellite signal blockage. To address these issues, we propose a Deep Neural Network-based localization system that combines passive Visible Light Positioning (p-VLP) and Bluetooth Low Energy (BLE) technologies to achieve stable, energy-efficient, and accurate indoor localization. Our solution leverages incremental learning to fuse data from visible light and BLE, overcoming their individual limitations and achieving centimeter-level localization accuracy. We build a prototype using low-cost S9706 hue sensors for p-VLP and low-power nrf52830 BLE boards to collect data simultaneously from both technologies in a 25m2 testbed. Our approach demonstrates a significant localization accuracy improvement of approximately 47% and 64% compared to individual p-VLP and BLE technologies, respectively, achieving a mean localization error of 20 cm.
7 August 2007 in International Journal of Network Management
The proliferation of different wireless technologies has led to fragmentation of interfaces and APIs that are used to control and manage air interfaces and wireless links. The emergence of new networking and application concepts that support increasingly heterogeneous access methods, especially when combined with future cognitive radio technologies, significantly increase the already high complexity of development projects and software architectures. Dissimilar wireless interfaces also complicate the development of middleware components for higher layers of the protocol stack. In this article we present a new generic interface and API architecture that has been developed and prototyped in several different wireless platforms to address these problems. We introduce the basic framework of the architecture, and explain how generic and open interfaces can significantly lower development and deployment costs. The developed interface architecture can be seen as a lower level middleware, or as an enabling technology to implement different technology transparent network middlewares. We also briefly comment about prototyping and standardization status of technology.
Exploitation of smart sensor data has the potential to significantly improve the way in which content is wirelessly delivered to end user devices within visitor attractions. Content delivery over wireless infrastructure can use knowledge and predictions of performance for intelligently prefetching and adaptively exploiting opportunities in order to enhance the overall user experience. The advantage of this approach over prior static multicast and on‐demand approaches is that it is possible to exploit location awareness (such as using global positioning system, iBeacons) combined with accelerometer and compass and other sensors, to plan and adapt content delivery in a more efficient and flexible manner. This study assesses different options for prioritisation and prefetching of content over dedicated wireless fidelity off‐load infrastructure to compare the ability to satisfy the user demands. The overall assessment, based on trials performed at two visitor attractions, shows that intelligent predictive prefetching of content can greatly reduce the need for expensive infrastructure compared with alternative static reliable multicast or on‐demand delivery strategies.