The first contribution of this task improves the functionality of network channel access and path selection protocols to support directed communication over the IEEE 802.11s network backbone. One or more network routers act as a network gateway and are connected to the Internet.
Motivation of the Research Work
In contrast, channel access requires information about the topology and interference characteristics of the forwarding path. The IEEE 802.11s network framework, which is a new standard in the wireless domain, is not specific to any of the above-mentioned works.
Contribution of the Thesis
The IEEE 802.11s MCCA-based channel access mechanism is used to control the channel reservation, so that each interface can reserve maximum channel based on solving the flow balancing problem. The optimal data rate is used for the peer establishment during the operation of the MPM protocol.
Organization of the Thesis
- Channel Access and Scheduling Optimizations for WMN
- Directional Antenna Support in Mesh Networks
- Service Differentiation and Fairness Provisioning in Mesh Networks . 19
- Design of Path Selection Protocols for Mesh Network
In [74], the authors showed that the capacity of a WMN depends on the implementation design. Most of the rate adaptation protocols proposed in the literature via the IEEE 802.11 technology are based on this concept.
IEEE 802.11s Standardization
- Enhanced Distributed Channel Access
- Mesh Coordinated Channel Access
- Mesh Path Selection Protocol in IEEE 802.11s
- On-Demand Reactive Mode
- Tree Based Proactive Mode
- Peer Management and Topology Control
The network STA that transmits an MCCA Configuration Request frame to initiate a reservation becomes the MCCAOP owner of the MCCAOP reservation. The MCCAOP owner and the MCCAOP responder advertise the MCCAOP reservation to their neighbors through MCCAOP advertisement messages.
Developments over IEEE 802.11s Standardization
Mesh peering Open frames are used to establish mesh peering between two neighboring mesh STAs. Although the scheme can improve network performance, it does not consider the impact of multiple interfaces over the mesh path selection decision.
Summary
In another work [180], the authors studied QoS-aware path selection based on the proactive and reactive HWMP elements. This chapter provides efficient probabilistic scheduling on top of MCCA channel access for efficient scheduling that minimizes interference between competing beams.
Motivation for Scheduling
It can also be observed that flow f4, due to the subflow f3(ST A4 → ST A3) and the subflow f1(ST A7 → ST A2), achieves a total transmission rate of 0.5, whereas it can achieve transmission rates up to 0 .66 if it can be planned correctly. This shows the requirement for a proper scheduling mechanism that should limit the transmission rates of all the subflows to the maximum achievable transmission rate of the corresponding end-to-end flow.
Network Model and Assumptions
However, due to the subflow of3(ST A4→ST A3), data packets may be congested at ST A3. Given the limitations of the number of channels, it is safe to assume that interfaces connected to a single STA network use separate, non-overlapping channels to avoid self-interference due to simultaneous transmissions and receptions.
Interference Model and Characterization
In the interference graph, there would be directed edges from 1 to 3 as well as 3 to 1. Accordingly, the receiver of the interface Ik can be at most two hops away from Ij.
Flow Scheduling for Directional Multi-interface WMN
- Centralized Formulation for Flow Scheduling
- Distributed Flow Scheduling
- Equilibrium Analysis
- Convergence Analysis
- Fairness Analysis
- Tuning MCCA for Scheduling based on Requirement
The solution of the centralized problem given in Problem 3.1 remains within the solutions of the LSP given in Problem 3.2. The subgradients of the LSP given in Problem 3.2 are also convex subgradients of Problem 3.1.
Mesh Path Selection based on Interface Scheduling
Limitations of HWMP
The next subsection provides a modification on HWMP based on the scheduling mechanism given in section 3.4 for efficient path selection in a directional mesh network.
HWMP Path Metric: Scheduling Based ALM
If a linkIs→It is maximally loaded, then based on the proposed scheduling policy, a newly introduced subflow Fnew(Is,It) can achieve at most ξ(Fnew,Is) amount of transmission rate. Further suppose that based on flow priority, the proposed scheduling mechanism allocates ξ(Fnew,Is) the amount of transmission rate initially for newly introduced flow Fnew.
Performance Analysis through Simulation
Simulation Setup
Based on this observation, the EDR at everyIs for a newly admitted current Fnew (EDRFnew(Is)) is calculated as follows: 3.14) The HWMP protocol is extended as follows. The proposed scheduling scheme is simulated with both ALM and S-ALM as the link metric for the HWMP protocol.
Analysis of the TCP Performance
The bundle prioritization method used in the proposed scheduling mechanism solves this problem by providing max-min fairness between the sub-flows of different flows based on their required transmission rate. 3.9, TCP fairness increases for the proposed scheduling along with S-ALM, compared to other schemes, as the number of flows in the network is increased.
Analysis of the UDP Performance
In the proposed scheduling mechanism, the packet delivery rate of a flow depends on the overall content of the network. The end-to-end delay for the UDP traffic versus the number of UDP sources is shown in Figure 2.
Summary
The centralized problem formulation for inter-class service differentiation and intra-class fairness is discussed in section 4.2. EDCA provides service differentiation by maintaining separate CW and AIFSN values for different service classes.
Centralized Problem Formulation for Service-Differentiation and Fairness . 69
- Distributed Convex Decomposition
- Sub-gradient Calculation
- Sub-gradient Projection
- Convergence and Equilibrium Analysis
- Tuning MCCA for Service-Differentiation
The local unconstrained optimization for the distributed decomposition given in Problem 4.2 can be represented as;. Then, according to Proposition 3 of [196], the following theorem can be directly derived for the convergence of the proposed scheme.
Call Admission Control using HWMP
- Working Procedure of CAC-HWMP
- Applicability of the Proposed Scheme for Four Class Service System 80
- Trade-off Between Proportional Fairness and Max-Min Fairness
- General Performance Parameters
Therefore, as the number of VO flows in the network increases, the throughput for the BE flows in the proposed scheme decreases. Therefore, increasing number of VO flows in the network does not affect the performance of BE flows.
Summary
The mesh architecture of IEEE 802.11s has been proposed as a queuing network to theoretically model the trade-off between rate-hop and interference. Based on this theoretical analysis, Section 5.3 proposes the speed adaptation in the IEEE 802.11s mesh network, HITRAM.
Queuing Analysis of IEEE 802.11s for a Specific Rate Region
Network Model
The proposed analysis can be extended for a multi-channel network by grouping network routers that use a similar channel (co-channel interference) or overlapping channels (adjacent channel interference). It can be observed that the service rate µi(r) may vary in different DTIM intervals, and without loss of generality, the analysis of the effect of multi-rate is limited to one DTIM interval.
Derivation of the Parameters for Queuing Network
The effect of the selected data rate on the number of hops and disturbances can be expressed by the following lemmas and theorems. For the data rate r, equation (5.19) shows the effect of the selected data rate on end-to-end delay based on the number of hops and interference information relative to the transmission range of the selected data rates.
Validation and Analysis of the Model
It can be seen from the figures that the results obtained from the theoretical model correspond to the simulation results for all three data rates. Low data rates can be chosen when the number of active mesh STAs in the neighborhood is less, to reduce the number of hops between end-to-end communication pairs.
HITRAM: Protocol Design and Implementation
- Mesh Beaconing
- Initial Peer Selection and Mesh Functioning
- Estimation of Rate-Hops-Interference Trade-off
- Selection of Optimal Data Rate
In this way, each STA in the network checks the possibility of using the low data rate to reduce the number of hops. If the rate is reduced, the MPM protocol sets peers based on the selected data rate.
Simulation Results
- Simulation Setup
- Performance for UDP Traffic
- Performance for TCP Traffic
- Compatibility with Other Rate Adaptations
- Impact of Different Channel Conditions
This series of simulation experiments investigates the impact of network load and number of customers on tariff adjustment schedules. The gradual performance degradation of HITRAM with a larger number of clients is only due to the bandwidth sharing between the TCP flows and the bottleneck behavior near the mesh gateways.
Summary
In a multi-hop mesh network, the variability in the channel and the network conditions also affect the mesh path selection protocols. The proactive path selection protocols decide the mesh path before the actual data communication takes place.
Network Model
The shortcomings of the proactive and reactive routing protocol, as mentioned earlier, also apply to the HWMP protocol. The goal is to find the best pair of interfaces for optimal network routing protocol performance.
Selective Greedy Forwarding (SelG): Protocol Design
Stage 1: Construction of the Set of Potential Forwarders
A set of PNCs is constructed using a proactive approach by receiving a PPREQ broadcast on various interfaces. It can be observed that the path metric values are stored together with the candidates in the PNC array.
Stage 2: The Greedy Selection of the Best Candidate
The utility function is designed based on the effect of the current network dynamics (channel state and interference) over stimetric information to find out the set of PNCs. Nevertheless, in practice, the numerical value of the bound O(|Fi| × |Γi|) is much smaller (maximum within the order of two digits).
Extension of the SelG Protocol for Path Selection Between Two
The Size of the Set of Potential Forwarders
Further assume that the path metric value for P1 is less than the path metric value for P2. Similarly, if the link is added to node B, the isotonicity of the path metric must also be satisfied.
Scenario Settings
Qualnet supports the IEEE 802.11s HWMP appeal protocol for MAC layer path selection in a complex network.
Mesh STA to Mesh Gate Communication
Therefore, variability in path conditions in terms of channel loss, contention, and network interference affects end-to-end packet delay. Therefore, the effect of channel and network variability has less impact on the end-to-end throughput performance.
Communication Between Two Mesh STAs
If the interference constraint is not met, the flows are temporarily redirected to one of the next best available paths until the interference constraint is met. Furthermore, the channel contention and interference also affect the performance of the reactive HWMP and SOAR protocols.
Summary
For this reason, the testbed is only used to analyze the performance of the proposed channel access and mesh path selection protocols. In the combined implementation of the proposed set of protocols over the mesh testbed, SelG uses S-ALM, as proposed in Chapter 3, instead of ALM.
Router Properties and Configuration
The performance data of the routers is collected for 6 hours and the average performance is shown in the graphs.
Evaluation of the Scheduling, Mesh Path Selection and QoS
Single Class Traffic: Performance Improvement
The average per flow throughput is calculated as the average throughput of all MAC layer flows in the network. The proposed planning and network path selection together with the S-ALM metric reduce the sudden drops of the fairness index by considering the actual planning information during the path selection procedure.
Multi Class Traffic: Inter-class Service Differentiation and Intra-
The S-ALM metric further improves fairness by selecting the best forwarding path for a new flow admitted to the network. Proportional fairness index reflects the service differentiation between the class while average Jain fairness index shows the fairness within the class.
Evaluation of the SelG Protocol: Mesh Path Selection Performance
Therefore, for the HWMP protocol, the minimum SINR value is observed in one of the routers in the forwarding path. Due to the reasons analyzed so far, SelG offers lower end-to-end latency than the HWMP protocol.
Summary
The SelG protocol provides better fairness between flows, compared to the HWMP protocol, as shown in Figure 7.11. Reactive HWMP distributes flows unevenly across the network and, as discussed previously, can result in network congestion, which is reflected in the fairness calculation.
Forwarding Load and Capacity Calculation of a Mesh Network
Communication and Interference Graphs
Communication Paradigms for Directional Antenna
Performance of Fixed Data Rates
Effect of Background Traffic
Structure of BEWARE Design
MCCAOP Reservation with Periodicity 2
Multiple Flow Scenario
Worst case interference scenario
Two Flow Scenario
Simulation Scenario
TCP Average Throughput
TCP Average Fairness Index
Average UDP Throughput
Average UDP Fairness
UDP Average Delay
UDP % Packet Loss
Throughput: VO Flows in Presence of BE Flows
Throughput: BE Flows in Presence of Voice Flows
Throughput: BE Flows in Presence of BE Flows
