No abstract

So far, the handoff management involved in the wireless local area network (WLAN) has mainly fallen into the handoff mechanism and the decision algorithm. The traditional handoff mechanism generates noticeable delays during the handoff process, resulting in discontinuity of service, which is more evident in dense WLANs. Inspired by software-defined networking (SDN), prior works put forward many seamless handoff mechanisms to ensure service continuity. With respect to the handoff decision algorithm, when to trigger handoff and which access point to reconnect to, however, are still tricky problems. In this paper, we first design a self-learning architecture applicable to the SDN-based WLAN frameworks. Along with it, we propose DCRQN, a novel handoff management scheme based on deep reinforcement learning, specifically deep Q-network. The proposed scheme enables the network to learn from actual users' behaviors and network status from scratch, adapting its learning in time-varying dense WLANs. Due to the temporal correlation property, the handoff decision is modeled as the Markov decision process (MDP). In the modeled MDP, the proposed scheme depends on the real-time network statistics at the time of decisions. Moreover, the convolutional neural network and the recurrent neural network are leveraged to extract fine-grained discriminative features. The numerical results through simulation demonstrate that DCRQN can effectively improve the data rate during the handoff process, outperforming the traditional handoff scheme.

No abstract

The growing physical (PHY) layer capabilities of Wi-Fi have made it possible to use Wi-Fi signals for both communication and human sensing. Wi-Fi channel state information (CSI) in PHY layer can be obtained from commodity Wi-Fi devices. As CSI can detect the minute environment changes that alter signal propagation, it is thus capable of capturing the subtle human activities to provide cost-effective and easy-to-use human sensing. However, the limited bandwidth of each individual Wi-Fi channel fundamentally constrains the resolution of signals, resulting in poor performance of human sensing. In this paper, we present WiRIM, a resolution improving mechanism for Wi-Fi based human sensing. We design a channel switching and aggregation algorithm to extend the effective bandwidth of commodity Wi-Fi signals and improve the performance and efficiency of human sensing applications. With aggregated CSI, WiRIM constructs feature images which contain rich frequency, temporal and spatial characteristics, and then uses a deep learning method to process the extracted features. We propose a cross-location human activity recognition (CLHAR) scenario as a case study. The CLHAR scenario requires a high enough resolution of the Wi-Fi signals to accurately recognize different activities under the interference of tiny changes in human location. The experiments demonstrate the generality and effectiveness of the proposed mechanism.

No abstract