2
Appendix C Route Discovery In this appendix, we comment briefly on the route discovery phase. In the model and analysis presented in Chapter 7, we have neglected this phase of communication, which, in reality, may have a significant impact on the overall delay experienced by a source node. More precisely, the delay in discovering a route should be added to the queuing delay for both RB (shown in Figure 7.3) and NRB (shown in Figure 7.5) schemes, respectively. Note that the route discovery phase refers to the phase that starts when a source node broadcasts a route discovery message until the moment when a source node knows how to reach the destination. In the RB scheme, it does not necessarily mean that the source node will get a reserved route immediately after it knows how to get to the destination. The conclusions drawn in Chapter 7 regarding the delay performance of RB and NRB schemes will not change provided that the delay in discovering a route is similar for both schemes. This happens, for example, if route discovery messages flood the network on a separate control channel in the same way for both switching schemes. Note that in an RB scheme, the nodes of a route discovered through the control channel may still be relaying messages for other sources, i.e. they may still belong to other active routes. In this case, the route will be activated only when all intermediate nodes will not be relaying any other flow (this time interval before activation is equal to the waiting time in the virtual request queue in Figure 7.3). In the NRB scheme, a source can use a control channel to discover a route as well: unlike in the RB scheme, however, as soon as a multi-hop route to the destination is found, the source can start transmitting (this is captured by the model in Figure 7.5). Therefore, one can partition the delays incurred by a source node in the two switching schemes as shown in Figure C.1. In both cases, the route discovery time, denoted as T d , is the same. In an NRB scheme, after discovering the route, it takes n h T NRB s to complete the message transmission, which can start immediately after route discovery. In an RB scheme, after discovering a route, the source has to wait W RB v s before activating it, and afterwards it takes T RB s more seconds to transmit the message. Let us denote the time that a node has to wait, before actually starting transmitting a generated message in the NRB and RB schemes, as T NRB tx and T RB tx , respectively. From Figure C.1, it is clear that T RB tx >T NRB tx . The final conclusions, in terms of delay, derived in Chapter 7 Ad Hoc Wireless Networks: A Communication-Theoretic Perspective Ozan K. Tonguz and Gianluigi Ferrari © 2006 John Wiley & Sons, Ltd. ISBN: 0-470-09110-X

Ad Hoc Wireless Networks (A Communication-Theoretic Perspective) || Appendix C: Route Discovery

Embed Size (px)

Citation preview

Page 1: Ad Hoc Wireless Networks (A Communication-Theoretic Perspective) || Appendix C: Route Discovery

Appendix C

Route Discovery

In this appendix, we comment briefly on the route discovery phase. In the model and analysispresented in Chapter 7, we have neglected this phase of communication, which, in reality,may have a significant impact on the overall delay experienced by a source node. Moreprecisely, the delay in discovering a route should be added to the queuing delay for bothRB (shown in Figure 7.3) and NRB (shown in Figure 7.5) schemes, respectively. Note thatthe route discovery phase refers to the phase that starts when a source node broadcasts a routediscovery message until the moment when a source node knows how to reach the destination.In the RB scheme, it does not necessarily mean that the source node will get a reserved routeimmediately after it knows how to get to the destination.

The conclusions drawn in Chapter 7 regarding the delay performance of RB and NRBschemes will not change provided that the delay in discovering a route is similar for bothschemes. This happens, for example, if route discovery messages flood the network on aseparate control channel in the same way for both switching schemes. Note that in an RBscheme, the nodes of a route discovered through the control channel may still be relayingmessages for other sources, i.e. they may still belong to other active routes. In this case, theroute will be activated only when all intermediate nodes will not be relaying any other flow(this time interval before activation is equal to the waiting time in the virtual request queue inFigure 7.3). In the NRB scheme, a source can use a control channel to discover a route as well:unlike in the RB scheme, however, as soon as a multi-hop route to the destination is found,the source can start transmitting (this is captured by the model in Figure 7.5). Therefore, onecan partition the delays incurred by a source node in the two switching schemes as shown inFigure C.1.

• In both cases, the route discovery time, denoted as Td, is the same.

• In an NRB scheme, after discovering the route, it takes nhTNRB s to complete the

message transmission, which can start immediately after route discovery.

• In an RB scheme, after discovering a route, the source has to wait WRBv s before

activating it, and afterwards it takes T RBs more seconds to transmit the message.

Let us denote the time that a node has to wait, before actually starting transmitting a generatedmessage in the NRB and RB schemes, as T NRB

tx and T RBtx , respectively. From Figure C.1, it

is clear that T RBtx > T NRB

tx . The final conclusions, in terms of delay, derived in Chapter 7

Ad Hoc Wireless Networks: A Communication-Theoretic Perspective Ozan K. Tonguz and Gianluigi Ferrari© 2006 John Wiley & Sons, Ltd. ISBN: 0-470-09110-X

Page 2: Ad Hoc Wireless Networks (A Communication-Theoretic Perspective) || Appendix C: Route Discovery

294 Appendix C. Route Discovery

Ttx = Td

Ts

Td

Wv

NRB Scheme RB Scheme

Ttx

nhT

NRB

RB

RB

RB

NRB

Figure C.1 Delay component of the two switching schemes

trendwise are therefore valid in an absolute sense. However, the real advantage of one schemeover the other may be relatively lower than the predictions of our analysis.