Re: 清华办的无线通信讲席教授组系列报告通知：牛人如织

I saw Ness Shroff was in the list! He's one of the big names here in OSU.

没空

本来想去听Tse的报告的，结果没时间啊。。

【 在 asin 的大作中提到: 】 : 本来想去听Tse的报告的，结果没时间啊。。 今天最后一天了 不过据说以后还有很多机会

I guess it should like this: Since INR/SNR -> infinite, the crosstalk dominates. So t1 -> r2 and t2 -> r1 are both error-free. Then r1 and r2 can exchange their received data, which leads to an 1 bit/s/Hz capacity. 【 在 heyjude 的大作中提到: 】 : 谁能解释一下，TSE今天下午讲的，INR/SNR--〉无穷的时候，为什么两用户的容量/速率 能达到1。 问了半天，还是不明白，很丢人的。

It should be like this: For Two User SISO case, when log INR/log SNR >= 2, the channel t1->r2 is much stronger than the channel t1->r1. When r2 treat its desired signal as noise, the achievable rate t1->r2 is still bigger than t1->r1 such that r2 can decode the message of t1. Since we are talking about a reliable decoding scheme, r2 of course can decode its desired message, therefore t1,t2->r2 form a MAC. Similar analysis can be obained at r1, thus each receiver can decode both messages at no rate loss. 【 在 ismydoom 的大作中提到: 】 : I guess it should like this: : Since INR/SNR -> infinite, the crosstalk dominates. So t1 -> r2 and t2 -> r1 are both error-free. Then r1 and r2 can exchange their received data, which leads to an 1 bit/s/Hz capacity.

嗯，受教了。想明白了。 关键是干扰和噪声本质上的信号结构不同，噪声是个实心儿的平台，而干扰只是一个空心儿的隔板。因此低于噪声平台的话死活也是过不去，但是低于干扰平台则可以钻过去。 so fancy!