Font Size: a A A

Research On Computation Offloading And Resource Allocation In Distributed Edge Networks

Posted on:2024-11-18Degree:DoctorType:Dissertation
Country:ChinaCandidate:W J HuangFull Text:PDF
GTID:1528307373971069Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
In recent years,driven by the rapid expansion of the Internet of Things(IoT)and its associated applications,there has been an exponential surge in the demand for highquality services.Due to the size limitation of IoT devices,they cannot meet the real-time requirements for computation-intensive or delay-sensitive applications.As a new computing paradigm,edge computing effectively mitigates the conflicts between the rapidly growing computing demand at the edge and the constrained computing capacity of IoT devices.In edge computing,IoT devices upload their computational tasks to edge servers with high computational capabilities for execution.The related results are returned to end users,thus significantly reducing computation latency and energy consumption.However,in massive heterogeneous edge networks,three key factors can lead to low quality of service: 1)In terms of edge service coverage,there is a contradiction between the limited communication coverage and the large-scale deployment of IoT systems,resulting in low coverage and poor quality of services.Although the existing multi-hop computation offloading approaches can improve edge service coverage,they often rely on the strong assumption of complete information acquisition,making them difficult to apply in realworld scenarios.2)In terms of task scheduling,distributed edge network systems usually contain multiple cross-domain,heterogeneous edge subsystems that compete with each other.The non-cooperative behaviors of selfish devices render traditional task scheduling algorithms,which are based on cooperative assumptions,ineffective.Furthermore,the mobility of terminal devices between multiple edge servers further reduces task scheduling efficiency.3)In terms of resource allocation,most of the existing work focused on transmission delay and energy consumption metrics,neglecting the constraints of data transmission deadlines.This results in insufficient utilization of channel resources.Some of the work considers the limitation of transmission deadlines,but uses ideal assumptions for the channel model;Moreover,the time-varying energy consumption of IoT devices,leads to unbalanced energy resources,increasing transmission delays.To solve these problems,in this paper,we investigate the computation offloading and resource allocation in massive heterogeneous edge networks,and the following three main research components are included.(1)Distributed multi-hop task offloading with incomplete information.Edge computing satisfies sudden demands of computation-intensive applications of Internet of Things(IoT)devices.Multi-hop task offloading has been a promising technology to provide edge services to areas with poor server coverage via multi-hop task forwarding.However,the existing multi-hop offloading approaches have primarily assumed that complete information is required,which does not always hold in heterogeneous IoT systems.To overcome this limitation,this doctoral dissertation proposes a two-stage multi-hop offloading algorithm aimed at minimizing the overall delay and energy consumption.Specifically,in the first stage,this paper proposes a hierarchical minority game to estimate the offloading cost based on the hierarchical estimation model and historical data.Each IoT device makes its own offloading decision by comparing the estimated offloading cost with the local computing cost.In the second stage,the tree-based routing mechanism schedules the efficient offloading transmission paths for the offloading IoT devices.Furthermore,the offloading delay is further reduced by building augmentation paths in the distributed tree structure.(2)Incentive-compatible mobility-aware offloading.To solve the problems of uncooperative behaviors of selfish nodes and network resource dynamics caused by the concurrent offloading of mobile devices,this paper proposes an incentive-compatible mobilityaware offloading strategy to maximize the system-wide profit.Specifically,an iterative hierarchical estimation algorithm is proposed to estimate the offloading delay and energy cost to iteratively optimize offloading decisions for IoT devices.An error compensation algorithm based on the long short-term memory network reduces the deviation of the estimated value from the true value.After that,an energy-efficient routing approach schedules the transmission paths for the offloading IoT devices.Furthermore,the proposed bidding policy incentivizes the relayers and calculators to engage in collaboration for system utility maximization.(3)Deadline-driven Adaptive Channel Allocation and Energy Management.The existing work on channel allocation often fails to establish channel schedules to meet the deadline requirement,as they often ignore the problem of packet loss and interference in wireless links.To solve this problem,this paper proposes a deadline-driven channel allocation scheme to maximize the packet delivery ratio before the deadline.Specifically,data transmissions are prioritized based on the link quality,deadline,and number of collisions to improve channel utilization.Furthermore,we propose a retransmission mechanism to efficiently utilize spare slots to accommodate potential retransmissions from lossy links.Moreover,we propose an adaptive energy management scheme to solve the problem of unsustained energy furnishings.A multi-layer collection tree is established to collect overall energy information,which can update the charging path based on the real-time collected energy information,increase the charging priority of low-energy nodes,and prevent them from dying.The above studies optimize distributed edge computing systems from three aspects:edge service coverage,distributed edge collaboration,and resource utilization.Experiments show that compared to the existing work,the proposed schemes 1)improves the service coverage of distributed edge computing systems and the quality of service;2)enhances system utility;and 3)reduces the waste of channel and energy resources,increasing the success rate of transmission delivery.The related work has been published in IEEE Transactions on Computers,IEEE Transactions on Mobile Computing and IEEE ICC.
Keywords/Search Tags:edge computing, Internet of Things(Io T), computation offloading, multi-hop routing, resource allocation
PDF Full Text Request
Related items