| 摘要: |
| 无线电频谱占用预测是认知无线电研究中的关键技术之一。实验采用中星世通CS-805F可搬移监测测向系统对四川省成都市的GSM900上行频段（890~915 MHz）和广播电视业务的部分频段（750~806 MHz）进行了为期24 h的实地监测，针对频谱监测中产生的大量历史数据，选用了部分周期模式的关联规则挖掘方法，挖掘频谱使用中存在的频繁模式，并由信道占用频繁模式生成强关联规则，得到特定业务频段的使用规律，从而实现无线电频谱的占用预测。实验结果表明，该方法在两个业务频段的占用预测均取得了较好的效果，准确率分别可达74.02%和83.98%。另外，实验指出了该算法的敏感参数并进行了简要分析。实验对研究认知无线电设备实施动态频谱接入和提高频谱使用率有一定意义。 |
| 关键词: 认知无线电 无线电监测；频谱预测 关联规则挖掘 部分周期模式 |
| DOI： |
|
| 基金项目:电子信息控制重点实验室基金项目（JS15120401535） |
|
| Wireless spectrum occupancy prediction based on association rule mining |
| MAN Fangwei,SHI Rong,HE Binbin |
| () |
| Abstract: |
| Wireless spectrum occupancy prediction is one of the key technologies in cognitive radio systems.A 24-hour spectrum measurement experiment is conducted in the uplink bands of GSM service(ranging from 890 MHz to 915 MHz) and the partial broadcasting services(ranging from 750 MHz to 806 MHz)，in Chengdu，Sichuan Province.For a large scale of history data produced by radio monitoring systems，association rules mining method in partial periodic pattern is chosen to mine frequent patterns in spectrum usage which can be used for generating association rules and acquiring using patterns of specific spectrum，thus realizing wireless spectrum occupancy prediction. The experiment results prove that the method can achieve a satisfactory prediction accuracy in two service bands(74.02% and 83.98% respectively).Moreover,the experiment points out sensitive parameters of this algorithm and offers a brief analysis of the parameters.The research has a certain significance for cognitive radio devices to apply dynamic spectrum access technology and improving spectrum utilization. |
| Key words: cognitive radio radio monitoring spectrum prediction association rule mining partial periodic pattern |
《电讯技术》编辑部 