Mobile video applications today have attracted significant attention. Deep learning model (e.g., deep neural network, DNN) compression is widely used to enable on-device inference for facilitating robust and private mobile video applications. The compressed DNN, however, is vulnerable to the agnostic data drift of the live video captured from the dynamically changing mobile scenarios. To combat the data drift, mobile ends rely on edge servers to continuously evolve and re-compress the DNN with freshly collected data. We design a framework, AdaEvo, that efficiently supports the resource-limited edge server handling mobile DNN evolution tasks from multiple mobile ends. The key goal of AdaEvo is to maximize the average quality of experience (QoE), i.e., the proportion of high-quality DNN service time to the entire life cycle, for all mobile ends. Specifically, it estimates the DNN accuracy drops at the mobile end without labels and performs a dedicated video frame sampling strategy to control the size of retraining data. In addition, it balances the limited computing and memory resources on the edge server and the competition between asynchronous tasks initiated by different mobile users. With an extensive evaluation of real-world videos from mobile scenarios and across four diverse mobile tasks, experimental results show that AdaEvo enables up to 34% accuracy improvement and 32% average QoE improvement.

In this study, mode transformation and united control mechanism for multiple subsystems in metro‐access optical network (MAON) are proposed to realise multi‐subsystem‐based virtual passive optical network. By combing the characteristics of multi‐point control protocols (MPCPs) of wavelength division multiplexing Ethernet passive optical network (WDM‐EPON) and orthogonal frequency division multiplexing passive optical network (OFDM‐PON), a unified MPCP (U‐MPCP) is proposed. Based on the U‐MPCP, flexible control node and central node are constructed to achieve asynchronous unified control of multiple subsystems. By designing new control frames, the mode transformation between two subsystems (WDM‐EPON and OFDM‐PON) can be carried out. It can not only finish the flexible transformation of optical network unit's (ONU's) working modes but also make the MAON have the backward compatibility to the existing access technologies. By creating transformation algorithm of ONU's working modes, the transmission delay of over‐load subsystem can be effectively reduced when ONU meets the condition of the mode transformation and finishes the transformation. Finally, by simulation, the effectiveness of the proposed scheme is demonstrated.