The concept of logical layer with reference to Two-Tier, BFMLM-FQ modules are discussed in the pervious chapters. The Hybrid Fair Packet scheduling Algorithm issues, described in this chapter can expand the logical layer further with an enhanced connectivity to explore the multilayered approach with the integration of the above referred two modules.
Ju.J Group  has carried out the problem with location- dependent contention in ad-hoc network design as shown in Fig. 8.1. Their model has not achieved the global properties in ad-hoc networks.
Thus the proposed Hybrid Fair Packet Scheduling Algorithm [54&55] ensures the optical ad-hoc network connectivity with like location-independent contention and global properties such as broadband connectivity and channel reutilization.
8.2 HYBRID FAIR QUEUING OPTICAL-AD-HOC NETWORK
The hybrid algorithm achieves the optical ad-hoc network design in three Phases. Phase 1 represents the conflicts between ensuring fairness and maximizing channel utilization. Phase 2 gives the node mobility and scalability. Phase 3 represents the throughput of fairness model by an integrated approach. Hence the global model can be achieved by hybrid fair queuing optical ad-hoc network architectures.
The following assumptions are made in the design:
1. Collision occurs when a receiver is in the reception range of two simultaneously transmitting nodes, thus unable to receive signals properly from either of them.
2. A node cannot transmit and receive packets simultaneously,
There are some more assumptions for the ad-hoc optical network global fairness design issues and are mentioned below.
* Flows are set to be infinite sources, i.e., each flow remains continually backlogged.
* The source traffic is considered to be of constant bit rate (CBR) for each flow.
* The physical channel capacity C is assumed to be one slot per time unit.
* Identical flow weight is used in order to compare the hybrid algorithm.
* The packet size per flow weight = 10.
* The throughput is counted as number of transmitted packets per unit time.
The respective flow charts for global fairness model and optical network throughput model are shown in fig. 8.2 and 8.3. The comparison table for the three models in logical layer is as shown in table 8.3.
8.2.1 LOCATION-DEPENDENT CONTENTION AND SPATIAL REUSE
The location-dependence allows more transmission flow and results spatial channel reuse in optical ad-hoc networks. Specifically, any two flows that are not interfering with each other can potentially transmit data packets over the physical channel simultaneously. The selection of simultaneous transmitters thus determines the aggregate channel utilization. Hence the packet scheduling discipline needs to perform a judicious selection of such simultaneous transmissions while taking into account fairness considerations across flows. Fluid fairness ensures local fairness in the time domain and global fairness in frequency domain in optical ad-hoc networks .
8.2.2 IMPLEMENTATION OF LOCAL FAIRNESS MODEL
The major difference between the global fairness model and the local fairness model is the definition of basic fair share for each individual flow. Otherwise, both models can use identical algorithms to improve spatial channel reuse, e.g., through simultaneously transmitting flows in the appropriate maximum independent set. This can be done by using the back off-based approach. Thus we need different approaches to realize the fair queuing algorithm in the basic channel for these two models. In this section, we focus on realizing the basic fair share using the local fairness model.
8.2.3 GLOBAL FAIRNESS MODEL IN HYBRID FAIR PACKET
The local fairness model implementation saves the overhead of maintaining the conflict-free information, and is more robust to dynamic topology in ad-hoc networks. However, this is achieved at the throughput rate of more piggybacked flow information in the data packets.
8.3 NUMERICAL RESULTS
The numerical results for Hybrid Fair-Queuing Scheduling Algorithm for 22 X 22 node connectivity is computed and presented in terms of input parameters like network connectivity, node connectivity and flow connectivity are as shown in tables 8.1 and the output parameters like throughput ratio, spatial channel reuse are as shown in table 8.2. By using network simulator (ns2) model the global fairness model of Hybrid Fair Queuing algorithm was achieved in terms of throughput and packet distribution in ad-hoc network connectivity as shown in Fig. 8.4.
In the previous work Ju.J Group et.al. model represented the Location-dependent method. This model suffered from achieving the global properties.
The effective throughput i.e. 83.9% for the local fairness model and 98.2% for the global fairness model. It results in higher aggregate throughput and higher spatial reuse in optical ad-hoc networks.
The previous work represented by Ju.J. Group analyzed ad-hoc network architecture by single channel connectivity only. A new model in optical ad-hoc networks fair packet scheduling is presented in which simulations are compared and integrated together to achieve the global fairness model, maximizing channel reutilization beginning with simple topologies with small number of nodes, and extended to a more complicated system with large number of nodes. For each example, a specific number of flows are presented first. Then the hybrid algorithm is applied and the global fairness model with maximum channel reutilization is achieved. Finally, the information is propagated by updating the tables at each node and its throughput is measured until the simulation comes to an end.
bf : flow f's backoff value in mini slots
zf : allocated transmission slots for flow F
rf : flow f's weight
F : the flow set in flow contention graph
Kf: the number of packet transmission flow in present flow
d(f) : degree of flow in the graph
N(f) : neighborhood flow