Jie Tian

With the deep integration of communication technology and Internet of Things technology, the edge network structure is becoming increasingly dense and heterogeneous. At the same time, in the edge network environment, characteristics such as wide-area differentiated services, decentralized deployment of computing and network resources, and highly dynamic network environment lead to the deployment of redundant or insufficient edge cache nodes, which restricts the efficiency of network service caching and resource allocation. In response to the above problems, research on the joint optimization of service caching and resources in the decentralized edge network scenario is carried out. Therefore, we have conducted research on the collaborative caching of training data among multiple edge nodes and optimized the number of collaborative caching nodes. Firstly, we use a multi-queue model to model the collaborative caching process. This model can be used to simulate the in-network cache replacement process on collaborative caching nodes. In this way, we can describe the data flow and storage changes during the caching process more clearly. Secondly, considering the limitation of storage space of edge nodes and the demand for training data within a training epoch, we propose a stochastic gradient descent algorithm to obtain the optimal number of caching nodes. This algorithm entirely takes into account the resource constraints in practical applications and provides an effective way to optimize the number of caching nodes. Finally, the simulation results clearly show that the optimized number of caching nodes can significantly improve the adequacy rate and hit rate of the training data, with the adequacy rate reaching 84% and the hit rate reaching 100%.

Decentralized Federated learning is a distributed edge intelligence framework by exchanging parameter updates instead of training data among participators, in order to retrain or fine-tune deep learning models for mobile intelligent applications. Considering the various topologies of edge networks in mobile internet, the impact of transmission delay of updates during model training is non-negligible for data-intensive intelligent applications on mobile devices, e.g., intelligent medical services, automated driving vehicles, etc.. To address this problem, we analyze the impact of delayed updates for decentralized federated learning, and provide a theoretical bound for these updates to achieve model convergence. Within the theoretical bound of updating period, the latest versions for the delayed updates are reused to continue aggregation, in case the model parameters from a specific neighbor are not collected or updated in time.

The accelerating progress of the Internet of Vehicles (IoV) has put forward a higher demand for distributed model training and data sharing in vehicular networks. Traditional centralized approaches are no longer applicable in the face of drivers’ concerns about data privacy, while Decentralized Federated Learning (DFL) provides new possibilities to address this issue. However, DFL still faces challenges regarding the non-IID data of passing vehicles. To tackle this challenge, a novel DFL framework, Hierarchical Decentralized Federated Learning (H-DFL), is proposed to achieve qualified distributed training among vehicles by considering data complementarity. We include vehicles, base stations, and data center servers in this framework. Firstly, a novel vehicle-clustering paradigm is designed to group passing vehicles based on the Bloom-filter-based compact representation of data complementarity. In this way, vehicles train their models based on local data, exchange model parameters in each group, and achieve a qualified local model without the interference of imbalanced data. On a higher level, a local model trained by each group is submitted to the data center to obtain a model covering global features. Base stations maintain the local models of different groups and judge whether the local models need to be updated according to the global model. The experimental results based on real-world data demonstrate that H-DFL dose not only reduces communication latency with different participants but also addresses the challenges of non-IID data in vehicles.

Advanced researches on connected vehicles have recently targeted to the integration of vehicle-to-everything (V2X) networks with Machine Learning (ML) tools and distributed decision making. Federated learning (FL) is emerging as a new paradigm to train machine learning (ML) models in distributed systems, including vehicles in V2X networks. Rather than sharing and uploading the training data to the server, the updating of model parameters (e.g., neural networks’ weights and biases) is applied by large populations of interconnected vehicles, acting as local learners. Despite these benefits, the limitation of existing approaches is the centralized optimization which relies on a server for aggregation and fusion of local parameters, leading to the drawback of a single point of failure and scaling issues for increasing V2X network size. Meanwhile, in intelligent transport scenarios, data collected from onboard sensors are redundant, which degrades the performance of aggregation. To tackle these problems, we explore a novel idea of decentralized data processing and introduce a federated learning framework for in-network vehicles, C-DFL( Consensus based Decentralized Federated Learning), to tackle federated learning on connected vehicles and improve learning quality. Extensive simulations have been implemented to evaluate the performance of C-DFL, that demonstrates C-DFL outperforms the performance of conventional methods in all cases.

Decentralized Federated learning is a distributed edge intelligence framework by exchanging parameter updates instead of training data among participators, in order to retrain or fine-tune deep learning models in a peer to peer way. Considering the various topologies of edge networks, the impact of transmission delay of updates during model training is non-negligible for data-intensive intelligent applications at edge, e.g., intelligent medical services, automated driving vehicles etc.. To address this problem, we analyze the impact of delayed updates for decentralized federated learning, and provide a theoretical bound for these updates to achieve model convergence. In addition, we propose novel decentralized federated learning (delay-adaptive-DFL) to enhance the corporate strategy of edge computing devices. It releases the requirement for aggregating the model updates of edge devices in a fixed time period. Within the theoretical bound of updating period, the latest versions for the delayed updates are reused to continue aggregation, in case that the model parameters from a specific neighbor are not collected or updated in time. Regarding four neural network models, we implement experiments on a real-world test bed, and demonstrate that delay-adaptive-DFLis more efficient than the latest baselines.

New intelligence applications are driving increasing interest in deploying deep neural networks (DNN) in a distributed way. To set up distributed deep learning involves alterations of a great number of the parameter configurations of network/edge devices and DNN models, which are crucial to achieve best performances. Simulations measure scalability of intelligence applications in the early stage, as well as to determine the effects of different configurations, thus highly desired. However, work on simulating the distributed intelligence environment is still in its infancy. The existing simulation frameworks, such as NS-3, etc., cannot extended in a straightforward way to support simulations of distributed learning. In this paper, we propose a novel discrete event simulator, sim4DistrDL, which includes a deep learning module and a network simulation module to facilitate simulation of DNN-based distributed applications. Specifically, we give the design and implementation of the proposed learning simulator and present an illustrative use case.

To enhance the quality and speed of data processing and protect the privacy and security of the data, edge computing has been extensively applied to support data-intensive intelligent processing services at edge. Among these data-intensive services, ensemble learning-based services can, in natural, leverage the distributed computation and storage resources at edge devices to achieve efficient data collection, processing, and analysis. Collaborative caching has been applied in edge computing to support services close to the data source, in order to take the limited resources at edge devices to support high-performance ensemble learning solutions. To achieve this goal, we propose an adaptive in-network collaborative caching scheme for ensemble learning at edge. First, an efficient data representation structure is proposed to record cached data among different nodes. In addition, we design a collaboration scheme to facilitate edge nodes to cache valuable data for local ensemble learning, by scheduling local caching according to a summarization of data representations from different edge nodes. Our extensive simulations demonstrate the high performance of the proposed collaborative caching scheme, which significantly reduces the learning latency and the transmission overhead.

Novel smart environments, such as smart home, smart city, and intelligent transportation, are driving increasing interest in deploying deep neural networks (DNN) in edge devices. Unfortunately, deploying DNN at resource-constrained edge devices poses a huge challenge. These workloads are computationally intensive. Moreover, the edge server-based approach may be affected by incidental factors, such as network jitters and conflicts, when multiple tasks are offloaded to the same device. A rational workload scheduling for smart environments is highly desired. In this work, we propose a Conflict-resilient Incremental Offloading of Deep Neural Networks at Edge (CIODE) for improving the efficiency of DNN inference in the edge smart environment. CIODE divides the DNN model into several partitions by layer and incrementally uploads them to local edge nodes. We design a waiting lock-based scheduling paradigm to choose edge devices for DNN layers to be offloaded. In detail, an advanced lock mechanism is proposed to handle concurrency conflicts. Real-world testbed-based experiments demonstrate that, compared with other state-of-the-art baselines, CIODE outperforms the DNN inference performance of these popular baselines by 20 to 70 and significantly improves the robustness under the insight of neighboring collaboration.

A Risk and Similarity Aware Application Recommender System