| Summary: | 博士 === 國立臺灣大學 === 資訊工程學研究所 === 100 === With the increasing of wireless in this dissertation, Radio spectrum is a kind of limited natural resource as well as water and crude oil. Its use is licensed and assigned by governmental agencies. However, according to the statistics of the Federal Communications Commission (FCC), temporal and geographical variations in the utilization of the assigned spectrum range from 15% to 85%. The temporally unused spectrum is referred to as spectrum hole or white space.
A new wireless technology called Cognitive Radio (CR) is introduced to reduce the waste of spectrum resources. CR system dynamically accesses the spectrums to improve overall spectrum usage of frequency bands. However, the access of unlicensed users to licensed spectrums causes interference to the licensed users on the bands.
Therefore, the key of CR design is to improving overall spectrum usage in the constraint of avoiding harmful interference to the licensed users. To achieve our CR design, we focus on the following challenges.
In the MAC layer, we have:
(1) To control CR users’ traffic and channel selection for avoiding interference to licensed users.
(2) To achieve fairness between heterogeneous users in different CR systems
In the TCP layer, we have:
(3) To analysis and solve the impact of CR link to TCP layer
In this dissertation, we develop efficient CR MAC and TCP protocols to improve spectrum utilization. To th first problem, we present two functions: RTS-CTS-CRTS handshaking and active traffic control mechanisms. The RTS-CTS-CRTS handshake mechanism can select a channel with better transmission quality. Then, the active traffic control mechanism is used to control the amount of DATA packets transmitted on the selected channel. We analysis the problem and solution by Markov chains and simulate by NS2.
To the second problem, we study two fairness issues: uncoordinated of PU-detection ability and uncoordinated of spectrum unit size. We propose several Markov-chain models to study the potential unfairness problems. The proposed solution is a jamming-based MAC-layer approach, called probing function, to enhance PU-detection ability and fairness feature. A MAC protocol, called CCR-MAC, is proposed based on the combination of probing function and our previous work on CR MAC to avoid hardware limitation and protect PU’s traffic at the same time.
To problem (3), we verify the ineffectiveness of WTCP in CR environment by NS2 simulation. Afterward, we analysis the cause of throughput decay conditions and concluding them into three events: PU-Interference Loss, Mild-Congestion and RTT-Variance events. To improve TCP throughput, three event handlers are proposed: PU-Interference handler, Faster-Recovery mechanism and RTT-Adjustment function. By applying the three handlers, a cognitive TCP that adapting CR environment, called CR-TCP, is proposed. The results show that CR-TCP improve throughput over 50% in both stable and varying environment while WTCP solutions only improve 10% to TCP-Reno.
|
|---|
