Research On Need-aware Resource Adaptive Schemes In Smart Identifier Network

With the rapid development of information and network technology,Internet users and emerging applications are growing in strength.In addition,Internet has gradually evolved from a packet-switched network to an integrated service network that hosts mass service information,such as multimedia communications,data transmission.However,the traditional TCP/IP network only provides a coarse-grained best-effort service,and does not carry need of users and services.it is difficult for the network to intelligently allocate the network resources according to the user's and service's needs,resulting in poor user experience,low resource utilization and bad service quality.Aiming to meet the urgent need for massive emerging needs and improve the circumstance of lacking in perception and adaption mechanisms,the dissertation presents intensive researches on the need-aware resource adaptive schemes based on SINET from the following two perspectives,i.e.how to perceive needs and how to utilize needs.Main contributions and innovations of the dissertation are as follows:1.Focusing on how to perceive needs,a need-aware framework of the future Internet architecture is proposed in order to realize flexible embedding,integration and extension for massive emerging needs.By discussing the necessity and challenges of need perception,the dissertation detailed describes the overall idea and design of the need-aware framework,summarizes the advantages of the framework,and illustrates its basic workflow.Finally,a prototype is built to realize the framework,and presents evaluation based on resource adaptation mechanism taking content size as an example.Numerical results testify the feasibility and effectiveness of the framework with obvious advantages in several major aspects.2.Focusing on "how to utilize user needs(UNs)and what the effect is on user experience in SINET",a UN-aware model for describing the process of service selection and a UN-aware adaptive transmission scheme is proposed.In order to analyze the effect by introducing user needs in SINET,the dissertation takes user needs for preference into consideration,and builds a service selection probabilistic model based on user preference to describe preference based service decision-making in terms of statistical probability distribution of service selections.For analysing the effects of user preference on service selection,a detailed theoretical derivation and Monte Carlo simulation are performed in different patterns respectively.Numerical simulation results show that user need for preference makes the process of service selection following a certain regularity.Utilizing user need turns to be practicable and effective for a user to obtain a relative preferred service.In order to perceive and utilize user needs,the dissertation takes user need for different level into consideration,and proposes UAAT scheme for adapting to user's personalized transmission needs in SINET.UAAT scheme presents an effective way to differentiate user level.Besides,UAAT introduces a new component-Level Manager(LM)which can combine user level together with bandwidth and allocate corresponding bandwidth for different level users depending on the interface condition and the incoming GET level.The simulation results show that UAAT has implemented a better performance than the original scheme in utilizing available bandwidth,which helps to achieve highly personalized service.3.Focusing on "how to utilize service needs and what the effect is on resource utilization in SINET",a deadline-aware scheme with low overhead(DSLO)for datacenter networks is proposed,aiming at reducing the bandwidth overhead incurred by unproductive flows.To achieve this,DSLO introduces two kinds of methods to predict unproductive flows apart from the adaptive process.Basic DSLO is a sender-driven strategy for predicting unproductive flows based on the Explicit Congestion Notification mechanism.Adaptive DSLO proposes a probabilistic model and two switch-driven adaptive strategies(a threshold-based strategy and a probability-based strategy)to predict unproductive flows more precisely.Simulation results show DSLO's effectiveness on reducing both the overhead and the ratio of flows whose deadlines are not met compared with some typical mechanisms for scheduling deadline-aware flows.4.Focusing on "how to utilize service needs and what the effect is on service quality in SINET",a flow-based deadline scheduling scheme for datacenter networks(FBDS)is proposed,aiming to alleviate the unsatisfactory effect in the condition of high fan-in degree.FBDS makes the unilateral deadline-aware flow transmission with priority transform into a compound centralized single-machine deadline-based flow scheduling decision.In addition,FBDS blocks the flow sets and postpones some flows with extra time until their deadlines to make room for the new arriving flows in order to improve the deadline meeting rate.The simulation results on flow completion time and deadline meeting rate reveal the potential of FBDS in terms of a considerable deadline-sensitive transport protocol for deadline-sensitive interactive services.