[Part one begins a discussion of the basic evolution of wireless networks and how position location could be considered from the networking point of view. Part two offers a general overview of wireless mobile ad hoc networks and sensor networks. Part three introduces some aspects of mobility that need to be considered to balance the capacity–coverage trade-off that mobility causes. Part four presents the concept of cognitive radio together with some recent results in the area of position location.]
5.3.3 Basics of Cross-Layering for Reconfigurable Networks
Traditionally, networks were designed with a layering architecture in mind,
where functions were separated and organized to offer services to upper layers and use services from lower layers. Open System Interconnection (OSI) is the standard that shows the hierarchy followed for this architecture.
In contrast, reconfigurable networks impose new challenges to network designers, administrators, and operators, since the definition of functions and operations in a layered architecture is not as clear as for wireline networks. For example, it is well known that the TCP protocol offers a connection-oriented service and works on an end-to-end basis, reacting to congestion and making the transmitter reduce or increase the packet rate according to measures of round-trip delay.
In general, the effects on round-trip delay were due to high levels of congestion through the paths that the information traveled, but in networks with wireless links, this might not be true since delay might be caused by packet buffering because of conditions such as outage or interference. With this, TCP would calculate that congestion was taking place when a very different phenomenon was present.
Another example would be the application of routing algorithms, since these depend on the construction of a topology that depends on the status of each individual link; thus, physical layer phenomena affect algorithms of other layers, which did not occur in wireline networks. For this reason, it is important to integrate these functions and operations so that they can interoperate in order to provide more efficient services. This is what we call cross-layering.
In general, the treatment of these effects that traverse several layers can be formulated through the use of objective functions that need to be optimized, using restrictions where each of the effects to be studied is introduced as a constraint. A good discussion of some of these issues, as well as the formulation of challenges and new paradigms within cross-layering, can be found in Van Der Schaar and Shankar , where adaptation techniques and optimization methods are introduced to take advantage of cross-layering. Issues concerning fairness and competition are also briefly discussed as new paradigms that should be considered in the cross-layering of reconfigurable networks.
Cross-layering topics appearing in the literature include, for example, the idea of using functions in the data link layer such as the automatic repeat request (ARQ) to improve physical-layer performance and vice versa ; these improvements are obtained by using new multiple access mechanisms in the data link layer that provide network diversity. They also show that the application of cross-layering allows throughput of almost one packet per slot and small delays over several traffic-load ranges. The disadvantage is the increased complexity of the receiver and greater power usage.
In Berry and Yeh , resource allocation is considered within the context of cross-layering; resources include the transmission power and rate of each user. Measures of QoS at the network level and physical-layer performance are obtained. They use a flat-fading Gaussian multiaccess channel with bandwidth W, noise density N0/2, fading level h, and rate r to obtain the minimum power required to transmit at rate r that is less than the channel capacity, as follows:
This approach is extended to the multiaccess case to solve for the powers of each user with the rates lying at the extreme points of the capacity region. In De et al. , cross-layering is used in the context of reconfigurable networks for the case of sensors using CDMA for multiple access. Grids and hexagonal topologies are compared using a uniformly distributed user population. The cross-layering is applied by studying the physical layer constraints, such as BER, on performance measures at the data link and network level, such as throughput.