Читать книгу Wireless Connectivity - Petar Popovski - Страница 15

For the purpose of this chapter, we define a communication channel to be the physical resource that is used for a wireless transmission. In that sense, in spoken communication, the channel is created by the audible vibrations that take place in air or even another sound-propagating medium. It is useful to note that the communication channel is not the whole physical medium with all the vibrations, since there are vibrations that cannot be registered by ear and thus do not carry useful audio information. Furthermore, spoken communication uses a single communication channel: one cannot switch to another channel, such as in a TV receiver, in order to listen to the desired speaker and avoid the undesired one.

As already stated above, our discussion will be limited to the case in which all nodes use a single communication channel. In reality that can be, for example, a certain frequency to which all the nodes are “tuned”. Here we use the term “frequency” as it is used in a common language for, say, a TV frequency. One may argue that Zoya and Yoshi can agree to one frequency, while Xia and Walt can agree to tune to another frequency and in this way they do not need to share the channel with the link Zoya–Yoshi. This is indeed possible and we will discuss it in later chapters, when we introduce the notion of separation in frequency. On the other hand, it is also true that Zoya and Yoshi should first use some communication channel to agree upon which frequency they will use for communication. This agreement is, again, metadata or control information, such that the corresponding channel is often denoted as a control channel and can be shared by multiple nodes to come to an agreement about the frequency. For example, if Zoya decides to communicate with Xia, then she knows that she should try to find Xia at the control channel and, upon contacting her, use the control channel to decide which channel/frequency they should both be tuned to in order to communicate the useful data. However, the control channel is a common, shared communication channel and therefore the question of how to share that channel to send metadata remains valid.

The communication model used in this chapter is called a collision model. This is because the central assumption of the model is that if two or more nodes transmit simultaneously, then the interference that they cause to each other is manifested as a collision at the receiver. Upon collision, the receiver does not manage to retrieve any data successfully. Another assumption in the model, not really related to the issue of collision, is that a node operates in a half-duplex manner and cannot receive while transmitting. Most of the wireless systems that we encounter today are not full-duplex, that is, do not transmit and receive simultaneously at the same frequency channel. However, although technologically more complex, it is also possible to have full-duplex operation. Therefore, throughout the chapter we will occasionally revise the half-duplex assumption and discuss the changes that the full-duplex can bring into the design of a specific protocol or algorithm.

The communication between the wireless nodes is based on data packets. A transmitting node is capable of sending bits per second (bps) such that a packet of duration contains bits. All packets have the same duration, unless stated otherwise. In the collision model, a packet is treated as the smallest, atomic unit of information, such that either the whole packet is received correctly or it is lost. In other words, it is not possible to receive only some bits of the packet correctly. A packet sent by Zoya to Yoshi is received correctly if:

1 Yoshi is in the communication range of Zoya such that the distance between them is less than m;

2 No other communication node that is within m of Yoshi transmits while Yoshi is receiving the packet from Zoya.

The first condition above indicates that each transmission is omnidirectional. Due to the basic property of reciprocity in electromagnetic/radio propagation (see Section 10.9), each reception in our model is also omnidirectional. From this it follows that Yoshi receives a signal as long it is sent from a distance less than m, regardless of the actual position. The ingredients of the collision model are illustrated in Figure 1.1. Specifically, Figure 1.1(a) depicts the data rate of an idealized single link as a function of the distance between two communicating nodes. An example communication scenario is depicted in Figure 1.1(b), where two nodes are connected by a line if the distance between them is less than , indicating the possibility of having a link between them. Figure 1.1(c) exemplifies a possible time evolution of a process of packet transmission in the framework of the collision model. The packet from Zoya to Yoshi is received correctly, while the packet is not, due to collision with the packet sent simultaneously by Xia. Note that, by treating a packet as an atomic unit of information, even a partial overlap of and causes packet loss. On the other hand, Walt is outside the range of Zoya, such that he can receive without being interfered with by .

Figure 1.1 Communication model used in this chapter, referred to as a collision model. (a) Simplified dependence between the data rate and the distance, denoting a communication range . (b) An example topology with possible wireless links among devices. The distance between two connected devices is at most . (c) Collision model at work for the topology in (b).

The assumptions of fixed-length packets and always-destructive collisions are weakening the analogy with a conversation. If we think to relate a packet to a spoken word, then not all words have the same length and missing some letters of a word may still not destroy its comprehensibility. In fact, the collision model is rather pessimistic. In reality, one expects a certain continuity in comprehensibility/correctness of a packet: if the packets and from Figure 1.1 have only a tiny overlap, then both would have to be received correctly by Yoshi. So, why are we not accounting for such a phenomenon and remain pessimistic about the collision? This is for pedagogical reasons in order to have a gradual path to system design and optimization. At the first step, make a system that works when every collision is identical and deemed destructive. In the next step, pose the question: what if not all collisions are identical? This leads to a refinement of the communication model by entering “inside the collision” and analyzing the different types of collision, which we will do in the later chapters. Notably, some types of collision will not be destructive and some collided packets can be received correctly, which sets the basis for optimizing the protocols further.