Transmission Control Protocol (TCP) is the most commonly used transport protocol on the Internet. All indications assure that it will be an integral part of the future Internetworks. In this chapter, the authors present how regular TCP, which was designed for wired networks, is not suitable for dynamic spectrum access networks. They develop an analytical model to estimate the TCP throughput of dynamic spectrum access networks. Dynamic spectrum access networks deal with opportunistic spectrum access leading to greater utilization of the spectrum. The extent of utilization depends on the primary user’s traffic and also on the way the spectrum is accessed by the primary and secondary users. The proposed model considers primary and secondary user traffic in estimating the TCP throughput by modeling the spectrum access using continuous-time Markov chains, thus providing more insight on effect of dynamic spectrum access on TCP performance than the existing models.
Top1. Introduction
Radio spectrum allocation has traditionally been controlled rigorously by regulatory authorities through licensing processes. All countries have their own regulatory bodies, though regional regulators do exist. In the U.S., these spectrum-related regulations are done by the Federal Communications Commission (FCC). Spectrum has long been allocated in a first-come, first-served manner. Recent measurements indicate a substantial under-utilization of radio spectrum which is a consequence of static spectrum assignment. According to the FCC (FCC, 2003), temporal and geographical variations in the utilization of the assigned spectrum range from 15% to 85% (Akyildiz, Lee, Vuran & Mohanty, 2006). If radios could somehow use a portion of this unutilized spectrum without causing interference, then, there would be more room to operate and exploit. Such an idea relates to a broader concept of Dynamic Spectrum Access and is depicted in Figure 1 as was shown in (Akyildiz, Lee, Vuran & Mohanty, 2006). This figure shows a three-dimensional model of radio communication in which a device communicates in a finite band of frequency (called a channel in general) for a certain period of time and using a prescribed amount of power (regulated by FCC). Among such bands/channels of operation, those channels which are unused for a certain period of time are considered vacant and are referred to as spectrum holes or spectrum opportunities1 (Zhao & Sadler, 2007).
Figure 1. Dynamic spectrum access concept
The key technology which enables radio devices to shift their frequency of operation on demand to utilize the spectrum opportunities is called Software-defined Radio (Mitola & Maguire, 2001). A software-defined radio system is a radio communication system where components that are typically implemented in hardware (e.g. mixers, filters, amplifiers, modulators or demodulators, etc.) are instead implemented using software on embedded computing devices or on personal computers (Dillinger, Madani & Alonistioti, 2003). A more ambitious goal is to have a wireless device that is smart enough to analyze the radio environment and decide for itself the best spectral band and protocol at the lowest level of power consumption. Such a device has been called “Cognitive Radio”. Cognitive radio represents a significant paradigm change in spectrum regulation and usage, from exclusive use by licensed users, also called primary users (PUs), to dynamic spectrum access by secondary users (SUs) and their coexistence with PUs. A network consisting of cognitive radios is called a Cognitive Radio Network (CRN). CRNs are opportunistic networks. The basic premise of these networks is that the owner of a licensed spectrum will not be using the spectrum always. Hence, this unused and licensed spectrum can be utilized on a non-interfering basis by other users who have a need for the same.
TCP is the most commonly used transport layer protocol used by most of the applications running on Internet. All indications assure that it will be an integral part of the future Internetworks. In a TCP connection, the transmitter uses an adaptive window based transmission strategy. The number of unacknowledged TCP segments cannot be more than the TCP window size (
) which can be adjusted by either the sender or the receiver based on the available buffers or the congestion. The transmitter, therefore, does not allow more than 
unacknowledged packets outstanding at any given time. By adaptively changing the size of congestion window, TCP can control the flow of data based on the level of congestion in the network while the level of congestion is estimated through indicators like delay or packet loss.