Conclusions 156

To make cognitive radios technology more practical and widely acceptable we conclude our findings in literature by providing possible solutions to a few of the challenges faced by CR. We believe that FCC and IEEE standards for CR, switching to digital TV, using spectrum usage databases, and deploying multi-band antennas in CR are some of the key areas that will make the CR technology widely being used in daily life.

Our research on existing CR MAC protocols enables us to identify the problems and limitations in the literature. We conclude that the major classification of CR MAC protocols is based on whether they use the control channel in the ISM band (GCCC) or use a local control channel (non-GCCC). We find that both the GCCC and non-GCCC based CR MAC protocols have advantages specific to their own class but on the other

- 157 - hand, both suffer from certain disadvantages. For example, the CR MAC protocols which use the GCCC as a control channel enjoy the benefits of being always available and license-free, but contrary to this, the GCCC based CR MAC protocols are more prone to congestion and security threats. The non-GCCC CR MAC protocols are resilient to security threats, but it is difficult and time consuming for the CR nodes to discover non-GCCC.

Any CR MAC protocol that needs to be developed must be equipped with the design features provided in Table 7.1. For example, a MAC protocol could be either centralized or decentralized, contention-based or coordination-based (contention free), equipped with a single transceiver or multi-transceiver, etc. We conclude our findings through a table in which we list the characteristics of twenty CR MAC protocols reported in the literature.

Table 7.1 Design Features of a CR MAC Protocol

Common Control Channel

 GCCC  Non-GCCC  Assumed  Hybrid (DDH-MAC) Direct Access  Contention based  Coordination based

Dynamic Spectrum Access

 Genetic  Game Theoretic Access Mechanism  Time slotted  Random  Hybrid Access Number of Radios  Single transceiver  Multiple Transceivers Synchronous Asynchronous

Overlay/Underlay Proactive/Reactive Centralized/Decentralized

We believe that in order to analyse the PU behaviour in a CR network accurately, a CR MAC protocol must be investigated in different possible network scenarios. We have proposed four network case-scenarios that deal with different possibilities of the PU claim on a control channel. Under an ideal network scenario, the secondary user scans the GCCC for a BF, and after reading the information, switches to the PCCH to dialogue the control information. In the worst network scenario, if a PU claim has been sensed, then, unlike other CR MAC protocols, secondary users do not have to re- negotiate the control information and can simply resume the conversation on a backup control channel. In this way, the pre-transmission time and other network overheads are

- 158 - significantly reduced. We compute the pre-transmission time for all network scenarios and discover that the average pre-transmission time of DDH-MAC is smaller when compared with other CR MAC protocols. We strongly argue that pre-transmission time plays a very important role in any CR MAC protocol. It is an overhead which each MAC protocol must aim to minimize in all possible ways.

Introducing the concept of more than one control channel, and enabling the nodes to exchange control information safely and efficiently, are novel ideas in developing the CR MAC protocols, which are our major contributions in this study. In this chapter, our other contributions are discussing and evaluating the pre-transmission time. We show that the pre-transmission time plays an important role and a smaller pre- transmission time helps to yield a higher throughput as nodes have to wait for less time before the actual transmission starts.

A system performance could not be validated until one or more of the following modelling techniques have been applied: i) analytical modelling; ii) implementation/simulation modelling; iii) combining both analytical and simulation modelling; and iv) testbeds. We have used first three techniques to observe the behaviour of our proposed MAC protocol. Different parameters such as throughput, PU interference probability, SUs‟ transmission opportunities to utilize available white spaces, and contention amongst CR nodes were evaluated. The framework achieved the desired results by ensuring that the network constitutes an exact CR network that has the potential to respond to external events and has the capability to adapt and reconfigure according to network scenario.

It is concluded that equipping each SU with two radios helps avoid the hidden terminal problem and also keeps nodes rapidly updated about any network change that occurs in the CR network. This task is accomplished by using both the radios simultaneously (one radio to scan and observe network activities on the control channel, and the other radio to transmit data). We have revealed main advantages of our scheme: i) more than one control channel is supplied; ii) two transceivers efficiently solve the hidden terminal problem in a multi-channel environment; iii) the control channel saturation problem is overcome. In particular, when a SU is exchanging the control information and the channel become unavailable, SUs simply switch to a backup control channel and thus avoid re-dialoguing the control information. We quantitatively identified the trade-off between the network aggregated throughput and

- 159 - the channel utilization, which provided us some useful guidelines to improve the QoS parameters in CR networks.