今年做的一个小demo,发布一下

60 views
Skip to first unread message

Lin HUANG

unread,
Jan 19, 2010, 5:16:42 AM1/19/10
to GNU Radio / USRP 中文论坛
大家好!
 
附件是Orange Labs Beijing今年在GSM频段做的一个小demo。感兴趣的朋友看看吧。
 
两点说明:
1、原本是想做成一个宽带系统插在GSM的空闲频率中的,可惜后来USRP2始终没有搞定。这是让我觉得很遗憾的一个地方。
2、之所以选GSM频段,是因为觉得对于运营商来讲,TV频段的共享目前还是不靠谱的事,靠谱的是自己手中已有的频率,所以就选择了GSM频段。
 
黄琳
newsletter for CR demo v1.2.pdf

High Mesh China

unread,
Jan 20, 2010, 8:10:33 PM1/20/10
to Lin HUANG, GNU Radio / USRP 中文论坛
恭喜恭喜。
里面用的是海曼无限的RFX900子板吗?

--
北京海曼无限信息技术有限公司
http://www.open-mesh.com.cn

2010/1/19 Lin HUANG <huangl...@gmail.com>
--
您收到此邮件是因为您订阅了 Google 网上论坛的“GNU Radio / USRP 中文论坛”论坛。
要向此网上论坛发帖,请发送电子邮件至 gnuradio...@googlegroups.com
要取消订阅此网上论坛,请发送电子邮件至 gnuradio-usrp-...@googlegroups.com
若有更多问题,请通过 http://groups.google.com/group/gnuradio-usrp-ch?hl=zh-CN 访问此网上论坛。


Lin HUANG

unread,
Jan 20, 2010, 9:12:16 PM1/20/10
to High Mesh China, GNU Radio / USRP 中文论坛
恩,用到了。挺好的,没什么问题。

2010/1/21 High Mesh China <high...@gmail.com>

Chao Dong

unread,
Jan 20, 2010, 9:17:53 PM1/20/10
to Lin HUANG, GNU Radio / USRP 中文论坛
不错,认知的概念炒了很多年,现在说实话都有点看不懂了,还是实际的系统有感觉!
能否解释下optimal idel spectrum的实际含义?
Chao DONG
Nanjing University
Nanjing Institute of Communication and Engineering
Nanjing, China, 210007
dch...@gmail.com

Lin HUANG

unread,
Jan 20, 2010, 10:44:44 PM1/20/10
to GNU Radio / USRP 中文论坛
视频太大,不能直接发附件。我放到论坛的文件区了


 
---------- Forwarded message ----------
From: Lin HUANG <huangl...@gmail.com>
Date: 2010/1/21
Subject: Re: [gnuradio-usrp-ch] 今年做的一个小demo,发布一下
To: Chao Dong <dch...@gmail.com>
Cc: GNU Radio / USRP 中文论坛 <gnuradio...@googlegroups.com>


呵呵,所谓optimal就是挑了一个空白最宽的地方,或者说是能量最低的地方(因为有时候有一些弱信号我们也把它当噪声来处理)。很小白的方法吧,不过我一直觉得简单的方法往往最可靠。去年我们搞了半天循环平稳检测,结果还是必须辅以能量检测,否则虚警概率很高。
 
奉送一个小录像吧,制作得不好,但是比起newsletter,从中可以看到更多东西。
 
顺便再贴两条别人问我的问题:
1. Q:It’s very interesting demostration. And I have a little concerns on why you choose GSM band.  How much spectrum hole in current GSM system? I appreciate If you can share some information about spectrum survey on GSM band.
A: For the choice of GSM band for demo, there are two reasons. Firstly our opinion is the GSM spectrum is controlled by operator but the opening of TV band is uncertain in near future from the regulator’s view. Another reason is a technical purpose. At the beginning we planned to implement a wideband OFDM system which can cover several holes by turning off part of subcarriers. But in the TV band, we found the spectrum is very sparse, it is not easy to implement a very wideband system to cover more than two holes.  Anyway, as we said in the newsletter, the TV band sharing is not excluded.
As for the spectrum survey, it is ongoing. We’ll let you know if we complete it (maybe in the 2nd half of this year.) For indoor scenario, as you can see from the figure, more than 80% GSM carriers are vacant.
 
2. Q:Another thing is how much the sensing sensitivity in this demo. How to define it?
A:For the sensing sensitivity, because GSM signal has obvious power spectrum compared with the noise. So we simply defined the average power of part of sub-bands with the lowest power as the background noise power (maybe including interference from other cells but we take it as noise.), then the threshold is this value plus an offset (e.g. 10dB) according to experiment results.
 
 
 
 
2010/1/21 Chao Dong <dch...@gmail.com>

High Mesh China

unread,
Jan 20, 2010, 11:25:19 PM1/20/10
to Lin HUANG, GNU Radio / USRP 中文论坛
谢谢!
咱们的论坛好像被盾了。至少我这里访问不了。
能把你的newsletter for CR demo v1.2.pdf上传到我们公司的QQ群吗?想让那里的哥们也分享一下。

--
北京海曼无限信息技术有限公司
http://www.open-mesh.com.cn

2010/1/21 Lin HUANG <huangl...@gmail.com>

Lin HUANG

unread,
Jan 21, 2010, 12:24:05 AM1/21/10
to High Mesh China, GNU Radio / USRP 中文论坛
你们的qq群好像不能共享文件了

Lin HUANG

unread,
Jan 21, 2010, 2:41:41 AM1/21/10
to Su Jinzhao, GNU Radio / USRP 中文论坛
CR节点是立即退出的,切换时延主要决定于重新扫描信道所花的时间。我们现在的版本为了使扫描准确度高一些,每个信道扫描时间比较长,总共有大约100个信道,而且还扫了3编,结果大约花了1分多钟吧。
 
但是你知道,能量检测这种算法,其实不用扫那么久也可以的。

2010/1/21 Su Jinzhao <suji...@gmail.com>
论坛“GNU Radio / USRP 中文论坛”确实上不去了,GMail的附件也不能直接查看了,不知道这些和Google闹着离开中国有没有关系?呵呵。。

想问下,Master和Slave Node在通信过程中,如果发现GSM信道要用了,CR节点切换信道的时延有多大?

2010/1/21 Lin HUANG <huangl...@gmail.com>



--
Su Jinzhao

Lin HUANG

unread,
Jan 21, 2010, 3:01:12 AM1/21/10
to Su Jinzhao, GNU Radio / USRP 中文论坛
视频流被中断了。我们做得不是很好,发端那个播放器(VLC)并不知道链路已经断了,所以就继续发,这部分视频就丢失了。

2010/1/21 Su Jinzhao <suji...@gmail.com>
CR节点立即退出的话,如果有正在传输的视频流,是立即中断传输,还是将视频流暂时缓存等找到合适的信道后再接着传输呢?


2010/1/21 Lin HUANG <huangl...@gmail.com>



--
Su Jinzhao

Chao Dong

unread,
Jan 21, 2010, 8:48:11 AM1/21/10
to Lin HUANG, Su Jinzhao, GNU Radio / USRP 中文论坛
还是关于optimal idle spectrum的问题,在实际中,GSM的频段的稳定性如何,我的意思是,扫描一遍后,确定的最宽的idle spectrum是否就是当前最宽的,如果全扫描结束后,所认定的spectrum被占用了呢?
貌似有人用最优停止理论做过这个问题。
Chao DONG
Nanjing University
Nanjing Institute of Communication and Engineering
Nanjing, China, 210007
dch...@gmail.com
----- Original Message -----
From: Lin HUANG
Sent: Thursday, January 21, 2010 4:01 PM
Subject: Re: [gnuradio-usrp-ch] 今年做的一个小demo,发布一下

Lin HUANG

unread,
Jan 24, 2010, 9:19:51 PM1/24/10
to Chao Dong, GNU Radio / USRP 中文论坛
我没有具体查过GSM协议和现网的配置。但是我观察的结果是,下行频段,好像每个载波都配置有控制信道(比如广播信道、同步信道),所以如果这个载波在使用,就会有明显的功率谱,即使当时没有人使用业务。但是这个谱的强度会随着业务的变化而跳动。
所以,扫描的结果基本上是稳定的,就是这个小区(加上附近的几个小区)配置的载波数量。
 
在全扫描结束以后,每次发射信号的时候都会检测一下频谱是否被占用,就跟Wifi一样,CSMA的。

2010/1/21 Chao Dong <dch...@gmail.com>
Reply all
Reply to author
Forward
0 new messages