Research On Interconnecting Technologies For Heterogeous Resource Pools In Super Base Station Systems

With the constant development of mobile internet,the Internet of Things,etc.,large-scale terminal accesses are required to connect to mobile communication networks.High requirements are proposed to improve the coverage and speed of mobile communication networks and mobile internet traffic tend to be diverse.If mobile communication networks construct more base stations with the stacking method based on the current traditional base station architecture to provide communication service,it will inevitably lead to high deployment cost,high-energy consumption,etc.Therefore,a centralized base station architecture is proposed.The super base station as a kind of typical centralized architecture representative separates the radio frequency and baseband function to achieve the centralized arrangement of the computing resource pooling,and shares and distributes resources through resource allocation technology.In the small-scale application scene,super base station architecture has been preliminarily verified.However,difficult system capacity expansion and high-energy consumption existing under the large-scale application cause failure in the commercial process.Interconnection capacity is defined as the number of accessible maximum Remote Radio Heads(RRH).The following two factors restrain system interconnection capacity:On the one hand,the baseband functions between RRH and BBU are completely separated.Much data needs to be passed back through the fronthaul network.Since both the fronthaul network capacity and the ability of RRH are restricted,the system's interconnection capacity reduces.On the other hand,the dynamic high-speed switch as the only channel connects the BBUand RRH pool,and its performance directly influences the interconnection capacity scale of the system.Upon breaking through system capacity,it is very important to boost the commercial process of the super base station by reducing the energy consumption of heterogeneous baseband units based on the centralized pooling scheduling characteristics.In allusion to the above problems,the interconnection of a heterogeneous resource pool is researched.Firstly,the functional split strategy is introduced to research the dynamic functional split strategy of baseband functions for the goal of maximizing the system's interconnection capacity on the condition that both the fronthaul network capacity and RRH ability are restricted.Secondly,the dynamic high-speed switching technology of large-capacity and low-latency radio frequency baseband is researched to improve the interconnection capacity scale of the system.Finally,on the basis of breaking through the interconnection capacity of the system,the task latency sensitivity and heterogeneity factor of computing resources are considered in the interconnection system with the system-level construction of a heterogeneous resource pool.Essentially,this is performed to research the dynamic distribution strategy of resources in heterogeneous baseband units and reduce the energy consumption of the system.Main contributions and research results are as follows:(1)The dynamic functional split strategy of baseband functions with latency sensitivity is proposed,and the RRH single node's demands on the fronthaul network capacity are reduced to improve the interconnection capacity scale of the system.Firstly,considering both the fronthaul network capacity and RRH ability are restricted,the interconnection capacity optimization model of the system is constructed.More specifically,if all baseband function tasks are processed in baseband units(BBUs),the number of RRHs accessing the system decreases and the system interconnection capacity will be reduced due to the fronthaul network capacity restriction.If all baseband function tasks are processed in RRH,the task latency can't be guaranteed due to the restriction of the RRH processing ability,where the number of RRH accessing the system decreases and the system interconnection capacity reduces.Secondly,the dynamic functional split strategy of baseband function is proposed based on genetic algorithm thoughts.Finally,the proposed strategy is compared with a fixed functional split strategy.The simulation results indicate that in the full load of a RRH,the interconnection capacity of the system corresponding to dynamic functional split strategy is 2.4 times that corresponding to fixed functional split strategy 8,and is also 1.4 times that corresponding to fixed functional split strategy 7-2.Besides,a kind of radio frequency resource-pool implementation scheme,which supports the dynamic functional split of baseband function,is proposed.(2)The dynamic high-speed switching technologies proposed,and the switch size is improved to increase the system interconnection capacity scale.Firstly,the characteristic of the network business flow in the mobile network is analyzed,and the traffic model meeting the actual business is introduced.Secondly,based on such a traffic model,the queuing structure is designed and the Data Priority Aware(DPA)algorithm is proposed.Finally,the switch prototype is built,and scheduling algorithm performance is evaluated from an interconnection capacity scale,latency,and throughput.The experimental results indicate that,on the condition of meeting communication latency,the maximum interconnection capacity scale that proposed the DPA algorithm meets reaches 58×58 and is superior to other algorithms.(3)The dynamic distribution strategy of resources in heterogeneous baseband units is proposed to reduce the energy consumption.Based on solving the capacity expansion,another key problem that the super base station system shall solve is how to utilize the centralized pooling scheduling characteristics for the reasonable distribution of heterogeneous computing resources in accordance with task computing needs and for reducing the energy consumption of heterogeneous baseband units.Currently,some resource allocation policies can be used for the dynamic allocation of computing resources in accordance with task demands.However,the current strategy doesn't consider the heterogeneity of computing resources,so it couldn't be directly applied in super base station architecture.Firstly,the interconnection system of a heterogeneous resource pool for a super base station is constructed.Secondly,latency sensitivity and heterogeneity of computing resources are considered comprehensively to model the energy consumption of heterogeneous baseband units,and the Heterogeneous First Fit Decreasing(HFFD)algorithm and the Hybrid Grey Wolf Optimization Algorithm enhanced with Levy Flight and Differential Evolution(HGWO-LFDE)are proposed.The simulation results indicate,compared with the energy consumption of traditional base stations,the energy consumption of the system corresponding to the HFFD algorithm and the HGWO-LFDE algorithm reduces significantly.In the low-load condition,compared with the energy consumption of a traditional base station,the HGWO-LFDE algorithm and the HFFD algorithm can save 88%-100%and 71%-100%energy consumption,respectively.In the high-load condition,compared with the energy consumption of a traditional base station,the HGWO-LFDE algorithm and the HFFD algorithm can save 88%-90%and 71%-82%energy consumption,respectively.In comprehensive consideration of two kinds of algorithm complexity,in actual application,the high-load time and low-load time shall apply the HGWO-LFDE algorithm and HFFD algorithm,respectively to obtain better system performance.Besides,the prototype of the super base station is proposed and it is foundation of the large-scale application.