arXiv subjects

Jingdong Xu

Publications and source records attributed to Jingdong Xu.

4 recordsLinked to original sources

arXiv2026

Heterogeneous Optically-Detected Spin-Acoustic Resonance in Solid-State Molecular Thin-film

We report an implementation of spin-acoustic resonance in pentacene thin films integrated on a high-quality-factor (high-Q) surface acoustic wave (SAW) resonator on a lithium niobate substrate. Heterogeneous optically detected spin-acoustic resonance (HODSAR) is an optically detected spin-resonance measurement in which the resonant drive is delivered mechanically by a surface acoustic wave (SAW). By leveraging the photo-excited triplet state of pentacene at room temperature, we demonstrate coherent spin manipulation via acoustic driving under zero externally applied magnetic field. The heterogeneously integrated device, referred to as HODSAR, utilizes spin-phonon coupling to achieve mechanically driven, zero-field spin resonance, opening avenues for room-temperature mechanically addressable spin control and device integration. We show that the high-Q multimode response of the SAW resonator enables spectrally selective acoustic addressing of triplet transitions near 105 MHz. Coherent control is evidenced by Rabi oscillations, with a Rabi frequency that increases linearly with the square root of the applied RF input power over the measured drive range, consistent with driven two-level dynamics under acoustic excitation. These results establish spin-acoustic resonance in a heterogeneously integrated molecular thin-film platform and provide a quantitative basis for benchmarking mechanically mediated spin control.

quant-ph↗

arXiv2023

When Computing Power Network Meets Distributed Machine Learning: An Efficient Federated Split Learning Framework

In this paper, we advocate CPN-FedSL, a novel and flexible Federated Split Learning (FedSL) framework over Computing Power Network (CPN). We build a dedicated model to capture the basic settings and learning characteristics (e.g., training flow, latency and convergence). Based on this model, we introduce Resource Usage Effectiveness (RUE), a novel performance metric integrating training utility with system cost, and formulate a multivariate scheduling problem that maxi?mizes RUE by comprehensively taking client admission, model partition, server selection, routing and bandwidth allocation into account (i.e., mixed-integer fractional programming). We design Refinery, an efficient approach that first linearizes the fractional objective and non-convex constraints, and then solves the transformed problem via a greedy based rounding algorithm in multiple iterations. Extensive evaluations corroborate that CPN-FedSL is superior to the standard and state-of-the-art learning frameworks (e.g., FedAvg and SplitFed), and besides Refinery is lightweight and significantly outperforms its variants and de facto heuristic methods under a variety of settings.

cs.NI↗

arXiv2021

Cost-efficient and Skew-aware Data Scheduling for Incremental Learning in 5G Network

To facilitate the emerging applications in 5G networks, mobile network operators will provide many network functions in terms of control and prediction. Recently, they have recognized the power of machine learning (ML) and started to explore its potential to facilitate those network functions. Nevertheless, the current ML models for network functions are often derived in an offline manner, which is inefficient due to the excessive overhead for transmitting a huge volume of dataset to remote ML training clouds and failing to provide the incremental learning capability for the continuous model updating. As an alternative solution, we propose Cocktail, an incremental learning framework within a reference 5G network architecture. To achieve cost efficiency while increasing trained model accuracy, an efficient online data scheduling policy is essential. To this end, we formulate an online data scheduling problem to optimize the framework cost while alleviating the data skew issue caused by the capacity heterogeneity of training workers from the long-term perspective. We exploit the stochastic gradient descent to devise an online asymptotically optimal algorithm, including two optimal policies based on novel graph constructions for skew-aware data collection and data training. Small-scale testbed and large-scale simulations validate the superior performance of our proposed framework.

cs.NI↗

arXiv2016

Content Retrieval At the Edge: A Social-aware and Named Data Cooperative Framework

Recent years with the popularity of mobile devices have witnessed an explosive growth of mobile multimedia contents which dominate more than 50\% of mobile data traffic. This significant growth poses a severe challenge for future cellular networks. As a promising approach to overcome the challenge, we advocate Content Retrieval At the Edge, a content-centric cooperative service paradigm via device-to-device (D2D) communications to reduce cellular traffic volume in mobile networks. By leveraging the Named Data Networking (NDN) principle, we propose sNDN, a social-aware named data framework to achieve efficient cooperative content retrieval. Specifically, sNDN introduces Friendship Circle by grouping a user with her close friends of both high mobility similarity and high content similarity. We construct NDN routing tables conditioned on Friendship Circle encounter frequency to navigate a content request and a content reply packet between Friendship Circles, and leverage social properties in Friendship Circle to search for the final target as inner-Friendship Circle routing. The evaluation results demonstrate that sNDN can save cellular capacity greatly and outperform other content retrieval schemes significantly.

cs.NI↗