基于AI算法的无线网络邻区关系优化研究与实践

针对现有邻区优化方式的不足,基于现网数据引入XGBoost机器学习回归预测算法,通过学习具有自动邻区关系网络的两两小区切换占比建立预测模型,优化非自动邻区关系网络小区邻区关系.研究结果表明,基于AI算法的无线网络邻区关系优化能有效提高邻区优化效率,提升邻区关系的准确性.     

用于智能天线系统的多小区空时联合处理算法

TD-SCDMA系统同频组网时,邻区用户同样会产生多址接入干扰,而传统的联合检测无法消除.为了抑制邻区干扰,对上行链路接收端提出了一种将联合检测扩展到多个小区并与智能天线相结合的多小区空时联合处理算法.该算法根据对邻区用户信道估计的结果选择其中的强干扰用户,并将邻区强干扰用户和本小区用户的复合信道冲激响应同时纳入空时联合处理的系统矩阵.仿真结果表明,在室内-室外和高速车载信道模型下,采用多小区空时联合处理算法的接收机解调性能有3～4.8 dB的提升.此外,讨论了邻区相关信息的获取、邻区强干扰用户的选择和运算复杂度等具体实现时需考虑的问题.多小区空时联合处理算法可有效消除邻区强干扰用户的干扰,从而大幅提高了同频组网时系统的性能.     

一种基于多元回归算法扇区间业务均衡画像模型研究

随着网络规模不断扩大，业务量持续提升，网络负荷不均衡问题越发尖锐，网络资源未得到充分利用，用户上网感知受到影响，因此寻求精细化业务均衡对提升用户满意度至关重要。旨在探索精准业务均衡的解决方案，通过研究扇区间重叠覆盖率、站间距及邻区间的切换内在关系，并对多维度数据进行数学建模，最终构建关于MRO重叠覆盖率、站间距、邻区切换比例等多元线性回归业务均衡画像模型，解决扇区间业务均衡无法精准定位与量化的问题。通过模型计算，试验结果表明，多元线性回归业务均衡画像模型有效提升扇区间业务不均衡的优化效率，显著减少高流量预警、高负荷待扩容小区，对精细化业务均衡优化具有重要指导意义。     

基于话务迁移的基站智能节能减排

基于网络话务模型变化、网络覆盖变化和用户行为的分析,实现动态、智能调整网络设置,在闲时把1800 MHz小区的话务迁移到900 MHz小区,关闭1800 MHz小区达到节能减排目的,同时,在关闭1800 MHz小区后通过调整邻区参数、TCH话务忙门限(半数率门限)、小区下最大PDCH比率门限等设置,保证网络性能稳定,达到保证网络质量稳定和节能效果明显,且不影响用户感知,真正达到降本增效。     

基于X2接口TD-LTE系统邻区自优化流程设计

邻区关系的自动建立和更新作为SON另一个重要方面,直接影响到网络的整体性能和切换性能.网优的重要工作之一就是邻区关系的优化.本文基于传统的邻区列表优化方法,通过对TD-LTE协议的研究,提出一种新的邻区列表优化方法.首先基于正常的通信过程,添加未知小区的IP,并建立X2接口;然后根据X2接口和S1接口的参数,提出一种算法来解决遗漏和添加的小区列表,并且包括解决PCI冲突及PCI不稳定.仿真结果表明,该算法能够自动完成邻区列表更新,解决邻区列表的不合理配置,并且能够提高切换成功率,减小小区掉话率.     

LTE网络基于MR的小区相关性分析在邻区优化中的运用

阐述了LTE网络中小区相关性的一种定义方法,即基于重叠覆盖的比例来标称相关性强度,从相关性强度来定义邻区的优先级,用来指导精细化邻区优化,同时也可以判断网络级别和小区级别的重叠覆盖比例,旨在能更好地提高用户体验。     

基于异频切换负载均衡算法研究与实现

在商用LTE基站系统中,小区之间存在业务量不均衡的情况,特别是热点区域,某些小区处于高负载状态,为了均衡小区的业务负载,引入了移动负载均衡(MLB)技术。选择合适用户(UE)切换到相邻低负载小区,从而有效地提高了整体网络资源效率和吞吐量,同时也能增强用户的性能体验。提出了一种以用户的RSRP和邻区PRB利用率为参数,基于UE业务量的异频切换负载均衡算法,选择出适量的UE切换到目标邻区中,从而达到网络负载均衡。根据实际基站调试结果表明,该算法能够快速精确地切换特定业务量的UE,明确了要卸载的业务量,并且提升了网络吞吐量和UE吞吐量。     

基于干扰源定位算法的网络结构优化方法研究

随着LTE网络负荷不断增大,网络结构问题日益突显,同频干扰日益严重,对用户感知造成了明显影响,因此网络结构专项优化在后期网络维护中越来越重要。传统的网络结构优化方法效率很低,解决一个问题小区需要调整多个邻区。在现网中,如果调整过多的小区,不仅会影响网络的正常运行,而且可能会导致更多新的网络结构问题。本文创新性的提出了一种基于问题区域主干扰源精准定位算法的LTE网络结构优化方法,快速精准定位网络结构问题区域以及区域内的主干扰源,通过调整主干扰源,使得区域内所有被干扰问题小区的指标同时得到明显提升,极大提高了优化质量和效率。     

GSM/TD-SCDMA多系统邻区自动规划研究

目前TD与GSM网络系统内外邻区关系的规划、核查都是通过耗时费力的人工方式来实现,所以通过对多系统邻区自动规划算法的研究,实现快速、准确的生成所需的邻区关系,可以大幅度提高2G/3G网络优化中的工作效率,节省人力,在实际应用中具有重要意义,本文就TD与GSM网络系统内外邻区自动规划算法及其应用进行详细的介绍.     

基于XDR与MDT数据实现VoLTE用户感知问题精准定位研究

在目前4G网络的VoLTE业务中,快速准确识别用户通话感知中的掉话、通话断续、单通等问题,并精准定位问题原因,是非常困难的事情。通过基于移动核心网的端到端XDR话单记录数据,结合无线侧MDT测量数据中的RSRP、RSRQ、SINR以及经纬度数据可实现快速高效的问题识别与问题原因定位,提升VoLTE用户业务感知问题的优化效率。     

Performance analysis of a cellular network with multiple classes of calls

In this paper, we study a cellular mobile communications network with multiple cells and multiple classes of calls. The different classes of calls have different call holding times and residence time distributions. We consider a protocol mechanism under which a blocked call in a cell is either disconnected from the network or is deemed as a handoff call in a neighboring cell. Under this protocol, we prove that the stationary distribution of this cellular mobile network has a product form. This allows us to derive explicit expressions for handoff rates of each class of calls from one cell to another and the disconnecting probabilities for each class of new and handoff calls. Our numerical results show how these measures depend on the mobility of the mobile terminals in each cell and on the numbers of reserved channels.     

Opposition Based Artificial Flora Algorithm for Load Balancing in LTE Network

In recent years, telecommunication is progressed due to the development of Long-Term Evolution (LTE) standards. This LTE network provides high-speed wireless communication for mobile devices and satisfies the requirements of customers from multi-cells. However, due to the congestion of mobile devices, each cell in the network may get overloaded. So, load balancing is the main challenge to the LTE network for reducing congestion or load in the cell. For load balancing, the optimal cell selection method is presented in this paper. Initially, the load factor of each cell is estimated. Then the load factor is compared with the predefined threshold load value. After the comparison, the heavy loaded cell handover the users to the optimal cell or cell with minimum load. This optimal cell is selected with the Opposition Based Artificial Flora (OAF) algorithm. Simulation results show that the proposed approach decreases 6% of the call blocking ratio (CBR) and 14% of Call Dropping Ratio (CDR) than the previous approaches.

     

4G蜂窝网中移动互联网IP电视服务的优化研究

4G蜂窝网的频谱分配效率难以负担日益增长的移动IPTV服务与大视频流数据,为此提出了蜂窝网支持移动IPTV服务的优化方案,在保持传统语音电话质量的前提下,提高了移动IPTV服务的性能.首先,为语音电话分配较高的频率分配优先级以维护语音电话的QoS(Quality of Service);然后,采用软频谱复用技术来增强频谱的效率,并降低相邻蜂窝之间的频谱干扰;最终,设计了动态频谱分配算法,对相邻蜂窝间边缘频带的频谱分配进行协调与优化,进一步地提高了频谱利用率.仿真结果显示,本频谱分配算法在不降低语音电话服务质量的前提下,明显地提高了移动IPTV服务的性能.     

LTE高负荷小区聚类优化方法研究

随着低资费套餐的快速普及,LTE高负荷问题日渐突出,高负荷小区整治成为网络运维的重要任务。本文提出LTE高负荷小区聚类优化方法,将问题小区聚类至五大类十三小类,并对各类问题提出针对性的优化处理建议,形成系统的高负荷小区整治方法。     

大话务并发场景下语音质量提升方法的研究

LTE大话务语数并发场景的特点是业务类型多、分优先级调度和资源受限,是现阶段的优化难点.本文结合浙江移动通信网络分布于资源分配现状,通过一些语音质量提升方法的实践,提出了一种基于语音编码的网络优化方案.结合压力测试经验,分析了语数感知拐点,并提出了相应的优化解决方案,有效降低了大话务小区分组丢失率和DTX比例,从而成功实现了语音质量提升.通过这种方式对浙江移动通信网络设备进行优化,经过测试,发现结果满足原定的提升语音质量的目标.     

小基站部署方案规划算法

随着移动网络的飞速发展,小基站在LTE网络建设中的地位越来越重要。结合实际建网经验提出一种最优小基站部署方案规划算法——关联层次分析法。该算法结合层次分析法和灰色关联分析法的思想,确定各部署方案和各指标参数之间的关系,并对各方案进行排序,最终得到最优小基站部署方案。仿真结果表明,关联层次分析法在投资收益均衡方面优于其他传统算法,并且建网以后的用户体验也有所提升。     

A Dynamic Reservation-Based Call Admission Control Algorithm for Wireless Networks Using the Concept of Influence Curve

In this paper a dynamic channel reservation and call admission control scheme is proposed to provide QoS guarantees in a mobile wireless network using the concept of influence curve. The basic idea behind the proposed scheme is that a moving user, in addition to its requirements in the current cell, exerts some influence on the channel allocation in neighboring cells. Such an influence is related to the moving pattern of the users and is calculated statistically. Furthermore we developed a general analytical model to calculate the corresponding blocking probabilities for wireless networks with multiple platforms, which removes the commonly used assumption that new calls and handoff calls have same channel holding time. The numerical results demonstrate that our scheme outperforms traditional channel reservation schemes and can effectively adapt to the real time network conditions.     

一种基于机器学习的LTE高铁故障识别方法与实例

为保证高铁LTE网络质量,提升客户感知,及时识别并解决高铁LTE网络故障至关重要.现有技术识别高铁故障主要依靠道路测试与网管后台数据,不仅耗时耗力,而且由于高铁专网多采用拉远与小区合并建设方式,故障具体点位和隐性故障难以定位.在此背景下,本文提出了一种基于大数据和机器学习,针对高铁网络特性的天馈故障识别方法,可有效提升...     

A distributed implementation of mobility load balancing with power control and cell reselection in 4G network

We consider a self‐organizing network (SON) capability of mobility load balancing in a 4G network, which determines the transmission power level for individual base stations and cell reselection for individual mobile stations such that the network‐wide load is minimized while satisfying the minimum signal‐to‐noise and interference ratio (SINR) requirement of individual users. Both centralized and distributed schemes are proposed. The centralized scheme is based on the greedy algorithm, serving as a performance bound to the distributed scheme. The distributed scheme is to solve the system‐wide optimization problem in the flat network model, i.e. no central control node. Furthermore, it requires relatively low inter‐cell exchange information among neighboring cells over an inter‐cell channel, e.g. X2 interface in the LTE network. The proposed design objective is to minimize the number of mobile users that do not satisfy the specified average throughput, while distributing the user traffic load as uniformly as possible among the neighboring cells. Our simulation results for a uniform user distribution demonstrate that the proposed scheme can achieve up to almost 80% of a load balancing gain that has been achieved by a greedy algorithm in the centralized optimization. Copyright © 2014 John Wiley & Sons, Ltd.