| With the widespread adoption of video sensing devices,the IoT has gradually evolved into a new branch-the Internet of Video Things(IoVT).The massive amount of video data obtained by IoVT provides enormous potential for diverse video applications,such as smart cities,remote healthcare,intelligent manufacturing,and more.Above IoVT applications require deep and comprehensive analysis of the video content.This requires IoVT devices with limited computing power to upload the captured videos to edge servers with sufficient computing capacity.As a bridge connecting IoVT devices and edge servers,the wireless video transmission mechanism is a critical component in ensuring that these applications are implemented successfully.However,the limitations of IoVT devices and the time-varying wireless transmission environment pose significant challenges to the design of existing wireless video transmission mechanisms.Existing digital video transmission mechanisms often use various methods to achieve high compression ratios in source coding schemes such as HEVC.This can result in significant computational overhead due to complex nonlinear operations such as intra-frame prediction and motion vector search.Due to limitations in power consumption and cost,IoVT devices have limited computing capacity,and computation latency becomes particularly evident.This cannot meet the real-time requirements of IoVT applications.In addition,digital video transmission mechanisms developed based on the source-channel separation theorem will face the "cliff effect" under time-varying fading channels.To address the aforementioned issues,this dissertation proposes applying the uncoded video transmission mechanism to the IoVT scenario.Specifically,the uncoded video transmission mechanism replaces the complex compression coding mechanism at the sender with a unified three dimensional discrete cosine transform(3D-DCT).A group of video frames collected will undergo decorrelation processing via 3D-DCT.Then,the device will perform amplitude scaling on the obtained coefficients and map them to the in-phase and quadrature components of physical layer symbols for transmission.Since the uncoded video transmission mechanism is linear overall,it can reduce the computational delay caused by compressed videos on IoVT devices and has excellent adaptive ability to time-varying channel quality,effectively eliminating the "cliff effect".Therefore,this design can well meet the requirements of the IoVT system.However,the data transmitted in the uncoded video transmission mechanism has different levels of importance and needs to be non-uniformly protected,which is different from the digital video transmission mechanism where the data has the same level of importance.Additionally,existing resource optimization schemes based on uncoded transmission mechanisms mainly focus on downlink unicast or broadcast scenarios,while IoVT mainly focuses on the uplink transmission scenario of multiple devices.The independent power allocation of each device makes it difficult to directly apply existing resource optimization schemes.Furthermore,since the uncoded video mechanism only uses 3D-DCT for compression,its transmission efficiency is relatively low.In situations where wireless resources are scarce,this mechanism may abandon the transmission of much coefficients,which will result in a sharp decline in video transmission quality.To address these challenges,this dissertation divides the research on uncoded video transmission resource optimization technology for IoVT into the following two parts:(1)This dissertation proposes a resource optimization mechanism for multi-device uncoded video uplink transmission for IoVT.Firstly,this dissertation establishes a transmission model for multiple IoVT devices uploading uncoded videos to a single base station and derives a video quality loss model.Subsequently,this dissertation introduces a joint power control and subcarrier allocation optimization framework,constructing a problem that maximizes video reconstruction quality subject to constraints on sending power and carrier resource allocation.Finally,based on the problem characteristics,this dissertation divides it into two parts:power allocation and carrier matching.For the power allocation subproblem,this dissertation derives a optimal closed-form solution using the Lagrange multiplier method for a given carrier assignment scheme.For the carrier matching subproblem,this dissertation models it as a one-to-one bilateral matching problem and propose a bilateral matching algorithm to solve it.Simulation results show that this optimization mechanism effectively improves video transmission quality.(2)This dissertation proposes a resource optimization mechanism for uncoded IoVT video transmission based on NOMA.To address the contradiction between insufficient bandwidth resources and large amount of data transmission in smart perception scenarios such as smart transportation,this dissertation introduces NOMA technology into the uncoded video transmission mechanism to solve the problem of wireless transmission resource shortage.The devices in the system are divided into near-distance device group and far-distance device group and the signals of both groups are superimposed for bandwidth utilization efficiency improvement.In order to deal with the interference generated by superposition,this dissertation optimizes the power allocation and design the coefficient chumks scheduling mechanism to minimize video decoding distortion.For the former,this dissertation introduces a block successive convex approximation algorithm(BSCA)to solve the power allocation optimization problem.For the latter,by constructing the coefficient chunks scheduling problem as a threesided matching problem among the far and near devices’ coefficient blocks and the carrier resource sets,this dissertation proposes a two-step matching algorithm for pairing construction and superposition matching to solve it.Simulation results show that this optimization mechanism improves the transmission efficiency and video transmission quality. |
PDF Full Text Request