Deep Learning Based Proactive Caching for Effective WSN‐Enabled Vision Applications

Abstract

Wireless Sensor Networks (WSNs) have a wide range of applications scenarios in computer vision, from pedestrian detection to robotic visual navigation. In response to the growing visual data services in WSNs, we propose a proactive caching strategy based on Stacked Sparse Autoencoder (SSAE) to predict content popularity (PCDS2AW). Firstly, based on Software Defined Network (SDN) and Network Function Virtualization (NFV) technologies, a distributed deep learning network SSAE is constructed in the sink nodes and control nodes of the WSN network. Then, the SSAE network structure parameters and network model parameters are optimized through training. The proactive cache strategy implementation procedure is divided into four steps. (1) The SDN controller is responsible for dynamically collecting user request data package information in the WSNs network. (2) The SSAEs predicts the packet popularity based on the SDN controller obtaining user request data. (3) The SDN controller generates a corresponding proactive cache strategy according to the popularity prediction result. (4) Implement the proactive caching strategy at the WSNs cache node. In the simulation, we compare the influence of spatiotemporal data on the SSAE network structure. Compared with the classic caching strategy Hash + LRU, Betw + LRU, and classic prediction algorithms SVM and BPNN, the proposed PCDS2AW proactive caching strategy can significantly improve WSN performance.

Optimizations for Energy Efficiency in Software-Defined Wireless Sensor Networks

“…With the evolution of wireless sensor networks over time, the mechanisms which ensure effective data transmission and longer sensor lifespan have been required to evolve as well to match the end-user applications’ expectations [ 16 , 17 ]. Improvements are currently being developed in enhancing the programmability of such networks.…”

“…As an improvement approach to the previous study, the current work attempts to reduce the communication that is being generated from the sensor level by employing caching techniques, thus maintaining the information locally and performing a few simple constant time procedures on the sensor data. Several caching techniques were experimented in WSNs [ 16 , 22 ]; however, the current work extends the studies by combining caching with content awareness in an SDN controlled environment.…”

“…The current work extends this approach by proactively delivering the data from the central component towards the application plane and improving the QoS and the QoE of the proposed system. [ 16 ] is a study which explores content caching to reduce the energy consumption at certain designated nodes; however, the current work differs in that the content caching takes place at every sensor node. Another enhancement provided by the current research is that it proposes the use of a single type of sensor in the sensors’ plane, therefore simplifying the system design and its deployment.…”

Representation and Reasoning of Three-Dimensional Spatial Relationships Based on R5DOS-Intersection Model Representation and Reasoning Based on R5DOS Model

“…In recent years, spatial reasoning has evolved rapidly, thanks to the emerging AI applications in image processing [9,10], computer vision [11,12], and model prediction [13]. However, most studies on spatial reasoning focus on the spatial relationships on two-dimensional (2D) planes rather than those in three-dimensional (3D) spaces.…”

File Size Distributions and Caching for Offloading

“…The increase of memory capacity coupled with the decrease of memory prices leads to increased amounts of memory in mobile devices, and to the possibility to add memory to Radio Access Network (RAN) base stations, to be used for femto [3], D2D [4], [5] or edge caching [6]- [13].…”