An Optimal Probabilistic Multiple-Access Scheme for Cognitive Radios
An Optimal Probabilistic Multiple-Access Scheme for Cognitive Radios
复制标题
DOI：
10.1109/tvt.2012.2200051
复制
发表时间：
2012-05
影响因子：
6.8
通讯作者：
Doha Hamza;S. Aïssa
中科院分区：
文献类型：
--
作者：
Doha Hamza;S. Aïssa
研究了一个主用户和次用户共享同一信道资源的时隙多址系统。SU在时隙的开始处感测信道。如果发现空闲，则以概率1发送。如果忙碌，则它以一定的访问概率进行传输，该访问概率是其队列长度和是否有新的分组到达的函数。两个用户，即，PU和SU通过采用截断信道反转功率控制方案以固定传输速率进行传输。我们考虑错误感知的情况。SU的目标是优化其传输调度策略，以最大限度地减少其平均发射功率和最大可容忍的主中断概率的PU的未命中检测的约束下，其排队延迟。我们考虑两个方案关于次要的传输错误的反应。在所谓的延迟敏感（DS）方案下，从队列中移除错误接收的分组以最小化延迟，而在延迟容忍（DT）方案下，所述分组被保持在缓冲器中并且被重传直到正确接收。使用后一种方案，存在缓冲器丢失的概率，该概率也被约束为低于某个指定值。我们还考虑了PU保持无限缓冲区来存储其数据包的情况。在后一种情况下，我们修改了SU访问方案，以保证PU队列的稳定性。我们表明，性能显着变化，如果考虑到一个主队列的现实情况。在所有情况下，虽然延迟最小化问题是非凸的，我们表明，访问策略可以有效地获得使用线性规划和网格搜索一个或两个参数。
We study a time-slotted multiple-access system with a primary user (PU) and a secondary user (SU) sharing the same channel resource. The SU senses the channel at the beginning of the slot. If found free, it transmits with probability 1. If busy, it transmits with a certain access probability that is a function of its queue length and whether it has a new packet arrival. Both users, i.e., the PU and the SU, transmit with a fixed transmission rate by employing a truncated channel inversion power control scheme. We consider the case of erroneous sensing. The goal of the SU is to optimize its transmission scheduling policy to minimize its queueing delay under constraints on its average transmit power and the maximum tolerable primary outage probability caused by the miss detection of the PU. We consider two schemes regarding the secondary's reaction to transmission errors. Under the so-called delay-sensitive (DS) scheme, the packet received in error is removed from the queue to minimize delay, whereas under the delay-tolerant (DT) scheme, the said packet is kept in the buffer and is retransmitted until correct reception. Using the latter scheme, there is a probability of buffer loss that is also constrained to be lower than a certain specified value. We also consider the case when the PU maintains an infinite buffer to store its packets. In the latter case, we modify the SU access scheme to guarantee the stability of the PU queue. We show that the performance significantly changes if the realistic situation of a primary queue is considered. In all cases, although the delay minimization problem is nonconvex, we show that the access policies can be efficiently obtained using linear programming and grid search over one or two parameters.