Article Preview
Top1. Introduction
In the mobile Internet era, mobile phones and other intelligent terminals are becoming more and more popular, and the adaptation of the wireless communications business is also increasing quickly. Traditionally, Spectrum allocation policy is to provide authorized users with a fixed spectrum which is exclusive to use (FCC, 2002). While this policy has been working well in the past few decades, the proliferation of wireless services in recent years has exposed the shortcomings of this policy, such as spectrum scarcity and a large number of licensed spectrum is not fully utilized in both time and space. This is also called Spectrum Hole or Spectrum Gap, which provides a good opportunity for wireless communication.
Subsequently, Dynamic Spectrum Access (DSA) is proposed to solve the problem of reasonable utilization of wireless spectrum resources. DSA is a new spectrum sharing paradigm which takes advantage of spectrum holes to ease the spectrum shortage problem and improve the spectrum utilization. Cognitive Radio (CR) technology is a promising paradigm for addressing the spectrum scarcity problem through efficient dynamic spectrum access (DSA) (Jararweh Y, 2015). In DSA networks, users are divided into two types: the Primary User (PU) and the Secondary User (SU). The Primary User can freely access the licensed spectrum band, while the Secondary Users can watch for their chances to access idle licensed bands as well as freely access the public spectrum band (IEEE, 2007).
DSA network is a wireless network which has been applied to cognitive radio technology. There are three kinds of DSA models: Interweave, Underlay and Overlay. Interweave DSA model is studied in this paper. DSA network can potentially be applied to complex network environment such as the disaster scene to improve reliability and service quality. The authorized users can also rent out their licensed spectrum band for trade.
TCP is the most commonly used transport control protocol. TCP’s congestion control mechanism includes four algorithms: slow start, congestion avoidance, fast retransmission and fast recovery (Jacobson V, 1990). In the slow start stage, TCP tries to find and utilize the available network bandwidth by doubling its congestion window in one RTT (Round Trip Time). When the congestion window exceeds a threshold, it will enter into congestion avoidance stage by increasing the congestion window by 1 packet in one RTT. When packet loss happens, it will fast retransmit the lost packet and reduce its rate in order to avoid congestion, that is the fast retransmission stage. Packet loss is detected by duplicated ACKs or ACK time out. If an ACK times out happens, it will reduce congestion window to 1 packet and then perform a slow start. Otherwise TCP will halve its congestion window, and perform a fast retransmission. This is called the Fast Recovery. From the above summary, we know that TCP assume every packet loss is due to congestion, and its bandwidth detection is by the aid of slow stat and congestion avoidance. It works well in wired networks, but this is not true in DSA networks because the packet loss may be caused by channel switch and the available bandwidth on each channel is significantly different.