SimDC: A High-Fidelity Device Simulation Platform for Device-Cloud Collaborative Computing
Ruiguang Pei 1, Junjie Wu 1, Dan Peng, Min Fang, Jianan Zhang, Zhihui Fu 2, Jun Wang 2 · 2025
Read SimDC: A High-Fidelity Device Simulation Platform for Device-Cloud Collaborative Computing by Ruiguang Pei 1, Junjie Wu 1, Dan Peng, Min Fang, Jianan Zhang, Zhihui Fu 2, Jun Wang 2 (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-22288.txt.
Introduction
Testing software that splits work between a phone and the cloud usually means buying racks of phones or faking them on servers, and both routes mislead. Ruiguang Pei and co-authors build a middle path in a 2025 paper: a simulation platform that mixes real phones into a server farm. Developers can spin up many virtual devices while a set of true devices reports how real hardware behaves.
A traffic controller steers results between devices and cloud under rules the user sets, mimicking delays and drops of real networks. Tests on a public data set show the platform reproducing behavior that pure server fakes miss. The code is public. For teams that cannot afford a wall of test phones, the hybrid rig offers a closer mirror of the field.
Catalog record
| CREATOR | Ruiguang Pei 1, Junjie Wu 1, Dan Peng, Min Fang, Jianan Zhang, Zhihui Fu 2, Jun Wang 2 |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 53 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-22288
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
SimDC: A High-Fidelity Device Simulation Platform for Device-Cloud Collaborative Computing Ruiguang Pei 1, Junjie Wu 1, Dan Peng, Min Fang, Jianan Zhang, Zhihui Fu 2, Jun Wang 2 Abstract The advent of edge intelligence and escalating concerns for data privacy protection have sparked a surge of interest in device-cloud collaborative computing. Large-scale device deployments to validate prototype solutions are often prohibitively expensive and practically challenging, resulting in a pronounced demand for simulation tools that can emulate real-world scenarios. However, existing simulators predominantly rely solely on high-performance servers to emulate edge computing devices, overlooking (1) the discrepancies between virtual computing units and actual heterogeneous computing devices and (2) the simulation of device behaviors in real-world environments. In this paper, we propose a high-fidelity device simulation platform, called SimDC , which uses a hybrid heterogeneous resource and integrates high-performance servers and physical mobile phones. Utilizing this platform, developers can simulate numerous devices for functional testing cost-effectively and capture precise operational responses from varied real devices. To simulate real behaviors of heterogeneous devices, we offer a configurable device behavior traffic controller that dispatches results on devices to the cloud using a user-defined operation strategy. Comprehensive experiments on the public dataset show the effectiveness of our simulation platform and its great potential for application. The code is available at https://github.com/opas-lab/olearning-sim . I Introduction Cross-device collaborative computing, such as federated learning (FL) [ 1 ] , has emerged as a promising solution to meet the increasing demand for low-latency, high-throughput, and privacy-preserving applications [ 2 , 3 , 4 , 5 ] . However, due to the large-scale deployment, heterogeneity of edge devices, and the diversity of their operational behaviors, the development and evaluation of device-cloud collaborative computing algorithms and applications face significant challenges [ 6 , 7 , 8 ] . Therefore, it is crucial to design and implement a high-fidelity device simulation platform for device-cloud collaborative computing, which provides a controlled and reproducible environment for real-world deployment. With this simulator, developers can quickly understand the operational effectiveness of their collaborative computing algorithms and applications under various conditions at a relatively low cost. Most current simulators [ 9 , 6 , 7 , 8 , 10 , 11 , 12 , 13 , 14 , 15 ] predominantly rely on high-performance servers. These systems frequently utilize multi-threading, multi-processing, and distributed computing techniques, creating separate processes to represent devices within device-cloud collaborative computing frameworks. This methodology is highly beneficial for addressing substantial challenges in large-scale device-cloud collaborative simulation scenarios. It allows extending the simulation scope to tens of thousands of devices without significant difficulties. Assuming sufficient computational resources are accessible, this strategy enables the efficient simulation and validation of device-cloud collaborative computing algorithms, providing essential experimental data and insights for analysis. However, there is still a gap between such simulation methods and real device-cloud collaborative computing scenarios. The challenges encountered in real-world contexts can be categorized into two main areas: 1. The discrepancy between virtual computing devices and actual edge devices. Although the scalability of virtual computing resources (based on high-performance servers) can meet the needs of numerous edge devices, they do not accurately emulate the real-time responses or operational burdens of physical devices, such as energy efficiency. Furthermore, the heterogeneity of edge devices (e.g., smartphones ranging from high to low configurations) introduces differences in communication efficacy and computational latency. These disparities are pivotal in implementing device-cloud collaborative computing applications on edge devices. For example, a device is permitted to participate in collaborative computing only if the user experience remains unaffected, this will ultimately impact the overall system and applications [ 16 , 17 ] . 2. The divergence between edge device behavior in simulations and the real world. The behavior of edge devices within device-cloud collaborative computing simulations is often underestimated, but in practical deployments, they significantly diverge due to variables like time zones and network connectivity. These discrepancies can considerably influence both the efficiency of collaborative computing and the stability of the system [ 18 , 19 ] . The ramifications of such impacts can be dissected into system-level and application-level concerns. For the former, the cloud infrastructure (e.g., aggregation services) may experience fluctuating access loads due to the uneven distribution of devices across different localities. For the latter, network inconsistencies and disconnections driven by diverse regional factors can compromise the algorithm’s ultimate accuracy. Recent advances [ 13 , 12 , 20 ] have highlighted the importance of accurately reflecting real-world scenarios in device-cloud collaborative computing simulations. Nevertheless, a truly holistic system that concurrently satisfies the demands of extensive simulations, authentic device emulation, device diversity, and accurate representation of edge device behaviors remains absent. Such a system would closely replicate real-world device-cloud collaborative computing settings in a comprehensive end-to-end simulation. In light of these considerations, we introduce a device-cloud collaborative computing simulation system tailored for real-world scenarios (denoted as SimDC), grounded on the following foundational principles: (1) Scalability : Acknowledging that real-world conditions frequently entail handling requests from millions, even tens of millions, of devices. (2) Heterogeneous Computing Devices : Our system integrates actual edge devices while also ensuring a diverse spectrum of device capabilities is represented. (3) Simulation of Edge Device Behavior : We incorporate the realistic behavior of edge devices, acknowledging its substantial influence on the overall performance of the application. To this end, we have implemented a hybrid heterogeneous resource framework, comprising: High-Performance Server Cluster , which is employed to cost-effectively simulate an extensive array of devices for functional validation, paralleling approaches from existing literature; and Physical Mobile Phone Cluster , consisting of Android mobile phones with varied configurations to capture authentic operational responses during algorithm training. In conjunction with this, we have developed an adept device resource coordination mechanism, tailored for our hybrid resource clusters, aimed at minimizing temporal costs while catering to the heterogeneous simulation demands of users. Further ensuring the fidelity of device behavior simulation, our system includes a bespoke device behavior traffic controller named DeviceFlow . As previously analyzed, the behaviors of devices primarily exert system pressure on cloud services or influence the overall outcomes. From the perspective of cloud services, device behaviors are mainly characterized by fluctuations in request traffic and disconnections. Based on this, DeviceFlow simulates device behaviors primarily by controlling the variations in network traffic from the edge to the cloud, in line with user-defined operational parameters. The major contributions of our paper are summarized as follows: • We propose a towards real-world scenarios device simulation platform for device-cloud collaboration computing, which employs a hybrid heterogeneous cluster design to support low-cost evaluation of algorithmic correctness and obtain real physical operating responses. Meanwhile, we design an efficient device allocation strategy to maximize resource utilization and minimize time costs. • We implement a novel programmable device behavior traffic controller, which provides a simulation of the varied behavior of heterogeneous edge devices in real-world environments according to user-specified operation specifications. • Extensive experiments indicate excellent scalability (easily scalable to support a large number of simulation devices), as well as high accuracy and efficiency of the designed simulation platform. II Related Work II-A Device-cloud Collaborative Computing Federated Learning (FL), as a typical device-cloud collaborative computing paradigm, has garnered considerable interest since its emergence due to its ability