Accelerating Wireless Federated Learning with Adaptive Scheduling over Heterogeneous Devices
2
Citation
40
Reference
10
Related Paper
Citation Trend
Abstract:
As the proliferation of sophisticated task models in 5G empowered digital twin, it yields significant demands on fast and accurate model training over resource-limited wireless networks. It is vital to investigate how to accelerate the training process based on the salient features of practical systems, including heterogeneous data distributions and system resources both across devices and over time. To study the non-trivial coupling between participating device selection and their appropriate training parameters, we first characterize the dependency of convergence performance bound on system parameters, i.e., statistical structure of local data, mini-batch size and gradient quantization level. Based on the theoretical analysis, a training efficiency optimization problem is formulated subject to heterogeneous communication and computation capabilities among devices. To realize online control of training parameters, we propose an adaptive batch-size assisted device scheduling strategy, which prioritizes the selection of devices that offer good data utility and dynamically adjust their mini-batch sizes and gradient quantization levels adapting to network conditions. Simulation results demonstrate our proposed strategy can effectively speed up the training process as compared with benchmark algorithms.Future generation wireless networks (FGWN) consist of heterogeneous networks managed by different operators. In heterogeneous network the provision of continuous service for mobile nodes is a main issue for FGWN. Handoff process enables a mobile node (MN) to provide such a facility by transferring an active call from one base station (BS) to another BS. The heterogeneous wireless networks (HWN) will support the vertical handoff mechanism. In this paper, we present an overview about the vertical handoff for heterogeneous wireless network and detail about multiple attribute decision making (MADM) based handoff decision to select a best network.
Heterogeneous wireless network
Vertical Handover
Cite
Citations (7)
This paper proposes a heterogeneous wireless network with handover in the application layer. The proposed network requires no upgrade of wireless infrastructure and mobile terminals to convert the present homogeneous networks to the proposed heterogeneous network. Only installing application programs on the content server side and the mobile terminal side is required. The performance of the proposed network has been evaluated in a field trial using a mobile broadband wireless access (MBWA) air interface with wide coverage and a wireless local area network (WLAN) air interface with high throughput. The field trial results show that the maximum value of the handover outage time is only 170ms. The proposed heterogeneous wireless network is promising since both high throughput and wide coverage area are attained by a combination of the proposed handover scheme with the present homogeneous wireless networks.
Field trial
Application layer
Cite
Citations (3)
s Now-adays many different types of networks communicate among themselves to form of heterogeneous wireless access technologies with diverse levels of performance has been envisioned to characterize the next-generation wireless networks. In heterogeneous wireless networks, handoff can be separated into two parts: horizontal handoff (HHO) and vertical handoff (VHO). VHO plays an important role in fulfilling seamless data transfer when mobile nodes cross wireless access networks with different link layer technologies. Current VHO algorithms mainly focus on when to trigger VHO to improve connection QoS but neglect the problem of how one can synthetically consider all currently available networks (homogeneous or heterogeneous) and choose the optimal network for HHO or VHO from all available candidates. In this paper, we present an analytical framework to evaluate VHO algorithms. This framework can be used to provide guidelines for the optimization of handoff in heterogeneous wireless networks. Subsequently, we extend the traditional hysteresis-based and dwelling-timer-based algorithms to support both VHO and HHO decisions and apply them to complex heterogeneous wireless environments. We refer to these enhanced algorithms as E-HY and E-DW, respectively. Based on the proposed analytical model, we provide a formalization definition of the handoff conditions in E-HY and E-DW and analyze their performance. Subsequently, we propose a novel general handoff decision algorithm GHO to trigger HHO and VHO in heterogeneous wireless networks at the appropriate time. Analysis shows that GHO can achieve better performance than E-HY and E-DW. Simulations validate the analytical results and verify that GHO outperforms traditional algorithms in terms of the matching ratio, TCP throughput, and UDP throughput.
Cite
Citations (2)
In the development of the wireless communication of the future, heterogeneous wireless network integration is a developing trend. Such an environment poses a significant challenge: how can we choose an appropriate Network Access Point (NAP) to achieve load balance and offer high quality communication in heterogeneous wireless handover decisions? We apply fuzzy logic in heterogeneous wireless network handover decisions and introduce the fuzzy normalization concept to make parameters for heterogeneous wireless network normalization. Using a simulation, we determine under what conditions our handover decision algorithm can balance network system loading better than the traditional fuzzy algorithms applied in heterogeneous wireless network handover decisions.
Heterogeneous wireless network
Vertical Handover
Wireless WAN
Normalization
Cite
Citations (0)
Heterogeneous network is a network connecting mobile station (MS) with different kinds of wireless network. It will provide an efficient network connection for mobile users. When considering various wireless access technologies, it's difficult to search out reliable criteria to pick out the most effective offered wireless heterogeneous network. During this paper, our aim is to develop an economical wireless network choice algorithm which is optimized using modified PSO. PSO algorithm is proposed for optimizing the heterogeneous wireless environment comprising universal mobile telecommunication system (UMTS) and long term evaluation (LTE). Also, a fitness operate is employed to keep up desired quality of services throughout the transmission between one networks to another.
Heterogeneous wireless network
Wireless WAN
UMTS frequency bands
Municipal wireless network
Base transceiver station
Multi-frequency network
Cite
Citations (3)
Network selection issue in heterogeneous wireless network is a hot topic, it also is critical to next generation network. In this paper, we sum up a lot of metrics of heterogeneous wireless network selection, and design a heterogeneous wireless network selection model based on bidding mechanism, simulation result present the good effect on heterogeneous wireless network selection model based on bidding mechanism, at the same time, user preference and the price policy of operations' company have an efficient impact on heterogeneous wireless network selection.
Heterogeneous wireless network
Wireless WAN
Municipal wireless network
Cite
Citations (0)
Heterogeneous wireless networks (HetNets) provide a means to increase network capacity by introducing small cells and adopting a layered architecture. HetNets can allocate resources flexibly through time sharing, on-off scheduling and cell range expansion/contraction. As shown in [1] such networks are constrained to a capacity limit in terms of the maximum number of downloads per second which can be achieved for a given offered traffic density. This capacity can be obtained as the solution to a continuous linear program (LP) (see [1]). Using the above results we investigate the scope for capacity extension in a single macro cell network. In particular it is shown that multiple pico cells per hot spot can significantly extend capacity but the gains can only be realized in conjunction with on-off scheduling of the picocells. We further show that scheduling according to a small number of on-off modes is sufficient to realise this gain. We also develop tight capacity bounds which can be used to investigate larger networks in cases where the size of the LP is too large for direct solution.
Macro
Proportionally fair
Cite
Citations (1)
In the development of the wireless communication of the future, heterogeneous wireless network integration is a developing trend. Such an environment poses a significant challenge: how can we choose an appropriate Network Access Point (NAP) to achieve load balance and offer high quality communication in heterogeneous wireless handover decisions? We apply fuzzy logic in heterogeneous wireless network handover decisions and introduce the fuzzy normalization concept to make parameters for heterogeneous wireless network normalization. Using a simulation, we determine under what conditions our handover decision algorithm can balance network system loading better than the traditional fuzzy algorithms applied in heterogeneous wireless network handover decisions.
Heterogeneous wireless network
Wireless WAN
Vertical Handover
Normalization
Cite
Citations (11)
This paper proposes a heterogeneous wireless communication network with soft handover in the application layer. The proposed network requires no upgrades of wireless infrastructure and mobile terminals to evolve the present homogeneous networks to the proposed heterogeneous network. Only installin
Soft handover
Heterogeneous wireless network
Wireless WAN
Air interface
Cite
Citations (3)
The wireless network environment in the future is going to include various heterogeneous wireless networks. In a heterogeneous wireless network environment, vertical handoff processes caused by the movement of users are inevitable. However, the research related to how to achieve seamless vertical handoff and improve QoS guarantees in today's heterogeneous wireless networks is still something to work on. Therefore, this paper proposed the Advanced Vertical Handoff Translation Center (Advanced VHTC) architecture for Wi-Fi/WiMAX heterogeneous wireless networks, aiming to improve the quality of services in heterogeneous wireless networks by means of packet formats translation between heterogeneous wireless networks and the MIH-based seamless vertical handoff mechanism, which cut down the time for vertical handoff and obtain high quality and persistent connection service. Lastly, with the brilliant results of a complete simulation, this paper will show the performance in terms of delay, jitter, packets loss and throughput to prove that the Advanced VHTC for next generation heterogeneous wireless networks can provide good QoS guarantees and obtain persistent transmission with high performance while performing seamless vertical handoff.
Heterogeneous wireless network
Wireless WAN
Packet loss
Vertical Handover
Cite
Citations (3)