Deriving Cell Load from RSRQ Measurements

Deriving Cell Load from RSRQ Measurements
复制标题

从 RSRQ 测量得出小区负载

DOI:
10.23919/tma.2018.8506494
复制
发表时间:
2018
期刊:
2018 Network Traffic Measurement and Analysis Conference (TMA)
影响因子:
--
通讯作者:
M. Rupp
M. Rupp
中科院分区:
--
文献类型:
--
作者:
Vaclav Raida;M. Lerch;P. Svoboda;M. Rupp

文献摘要

被引文献

相似文献

无线系统的性能往往受到干扰的限制。在LTE中,参数RSRQ与系统干扰有关。可靠而合理地测量此参数可以估计当前小区负载的当前水平以及当前小区中的干扰,从而使我们能够使用众包的性能数据进行网络基准测试。然而,RSRQ的解释并不简单。我们指出,如果在测量设备的下行吞吐量为零的情况下测量RSRQ,则可以使用RSRQ来估计其他用户造成的小区负载。在这种情况下,我们预计RSRQ和可实现的吞吐量之间存在正相关,这一点我们通过现场LTE网络中的测量进行了确认。相反,我们表明,如果测量设备正在下载数据,则可以生成范围广泛的不同RSRQ值。作为一种极端情况,我们给出了RSRQ和吞吐量之间具有强烈负相关性的测量结果。网络监控应用程序的源代码通常是专有的,因此我们不知道RSRQ样本是a)在零下行链路吞吐量时采集的,b)在下行链路吞吐量测试期间收集的,还是c)两者的组合。在a)情况下,RSRQ为我们提供了关于细胞负载的宝贵的额外知识。在b)和c)的情况下,如果我们不能过滤出对应于非零下行链路吞吐量的样本,那么它仅仅是无用的。
The performance of wireless systems is often interference-limited. In LTE, the parameter RSRQ is connected to the system interference. A solid and sound measurement of this parameter allows for an estimation of the current level of cell load as well as interference in the current cell, enabling us to use crowdsourced performance data for network benchmarking. However, RSRQ is not straightforward to interpret. We point out that RSRQ can be used to estimate the cell load caused by other users if it is measured at zero downlink throughput of the measuring device. In such a case we expect a positive correlation between RSRQ and achievable throughput which we confirm by measurements in a live LTE network. Conversely, we show that if the measuring device is downloading data, a wide range of different RSRQ values can be generated. As an extreme case we present measurements with strong negative correlation between RSRQ and throughput. The source codes of the network monitoring applications are often proprietary, we thus do not know if RSRQ samples are a) collected at zero downlink throughputs, b) during a downlink throughput test or c) a combination of both. In case a) RSRQ provides us precious additional knowledge about the cell load. In cases b) and c) it is merely useless if we cannot filter out the samples corresponding to nonzero downlink throughput.