1. QoS in Mobile Ad Hoc Networks

2. Introduction
Mobile ad hoc networks (MANETs) are infrastructureless and intercommunicate using single-hop and multi-hop paths
Nodes act both as hosts and routers
Topology changes could occur randomly, rapidly, and frequently
Routing paths are created and deleted due to the nodal mobility

3. Applications of MANETs
Collaborative computing
Communications within buildings, organizations, ad hoc conferences
Communications in battlefields and disaster recovery areas
Sensor networks

4. Quality of Service (QoS)
QoS: A set of service requirements that are met by the network while transferring a packet stream from a source to a destination
QoS metrics could be defined in terms of one or a set of parameters
Examples: delay, bandwidth, packet loss, delay-jitter, etc.

5. QoS in MANETs
The use of QoS-aware applications are evolving in the wireless environments
Resource limitations and variations adds to the need for QoS provisioning
Use of MANETs in critical and delay sensitive applications demands service differentiation

6. Issues and Difficulties
Unpredictable link properties
Node mobility
Limited battery life
Hidden terminal problem
Exposed terminal problem
Route maintenance
Security

7. Compromising Principles
Soft QoS
After the connection set-up, there may exist transient periods of time when QoS specification is not honored
The level QoS satisfaction is quantified by the fraction of total disruption
QoS Adaptation
As available resources change, the network can readjust allocations within the reservation range (dynamic QoS)
Applications can also adapt to the re-allocations

8. QoS Support in Physical Channels
Since wireless channel is time varying, the SNR in channels fluctuates with time
Adaptive modulation which can tune many possible parameters according to current channel state is necessary to derive better performance
Major challenge: channel estimation – accurate channel estimation at the receiver and then the reliable feedback to the transmitter
Wireless channel coding needs to address the problems introduced by channel or multipath fading and mobility
Cross-layer issue: Joint source-channel coding takes both source characteristics and channel conditions into account

9. QoS Provisioning at the MAC Layer
For providing QoS guarantee for real-time traffic support in wireless networks, several MAC protocols based on centralized control have been proposed
For multihop networks:
The MAC protocol must be distributed in nature
It should solve the hidden and exposed terminal problems

10. IEEE 802.11 DCF IEEE 802.11 is a CSMA/CA protocol
In the distributed control function (DCF) mode:
After the node has sensed the medium to be idle for a time period longer than distributed inter-frame space (DIFS), it begins transmitting
Otherwise the node differs transmitting and backs off
When the medium becomes idle for a period longer than DIFS, the backoff timer is decremented periodically. The node starts transmission as soon as the timer expires
To reduce collisions, the sender and the receiver exchange RTS and CTS packets

11. QoS Support using IEEE 802.11 DCF
IEEE DCF is a best-effort type control algorithm
The duration of backoff is decided by a random number between 0 and the contention window (CW).
Service differentiation can be achieved by using different values of CW
When packets collide, the ones with smaller CW is more likely to occupy the medium earlier

12. Black Burst Contention Scheme
Nodes with best-effort traffic and nodes with real-time traffic use different inter-frame space values
When the medium remains idle long enough, right before sending their packets, nodes with real-time packets first contend for transmission right by jamming the media with pulses of energy, called BBs
Each contending node uses a BB of different length. The number of slots that forms a BB is an increasing function of the contention delay experienced by the node
Following each BB transmission, a node senses the channel for an observation interval
Since distinct nodes contend for BBs of different length, each node can determine without ambiguity whether its BB is of longest duration

13. MACA/PR
Multihop Access Collision Avoidance with Piggyback Reservation provides guaranteed bandwidth support for real-time traffic
The first packet in a real-time stream uses RTS/CTS dialogs to make reservations in the path
The sender schedules the next transmission after the current data transmission and piggybacks the reservation in the current data packet
Upon receiving the data packet correctly, the receiver updates its reservation table and sends an ACK
ACK serves for the renewal of reservation, not for recovering from packet losses

14. QoS-aware Routing at the Network Layer
Types of MANET routing protocols:
Proactive, table-based routing schemes
Reactive, on-demand routing schemes
Constraint-based routing schemes
These algorithms are based on the discovery of shortest paths
QoS-aware routing protocol should find a path that satisfies the QoS requirements in the path from source to the destination

15. CEDAR
Core Extraction Distributed Ad hoc Routing scheme dynamically establishes the core of the network, and then incrementally propagates the link states of stable high-bandwidth links to the core nodes
The route computation is on demand basis
Components of CEDAR
Core extraction
Link-state propagation
Route computation

16. Integrating QoS in Flooding-Based Route Discovery
Ticket-based probing algorithm
During the QoS-satisfying path search, each probing message is provided a limited number of tickets to reduce the scope of flooding
When one or more probes arrive at the destination, the path and delay/bandwidth information is used to perform reservation for the QoS-satisfying path
A simple imprecise model is used for the algorithm

17. PANDA Approach
Positional Attributes based Next hop Determination Approach (PANDA) discriminates the next hop based on the desired QoS metric
Instead of using a random rebroadcast delay, the receiver opts for a delay proportional to its ability in meeting the QoS demands
The decisions at the receivers are made based on a predetermined set of thresholds

18. QoS Support using Bandwidth Calculations
The end-to-end bandwidth can be calculated and allocated during the admission control phase
Using TDMA, time is divided into slots, which in turn are grouped into frames
Each frame contains two phases: control and data.
During the control phase, each node takes turns to broadcast its information to all the neighbors in a predetermined slot.
At the end of control phase, each node knows about the free slots between itself and its neighbors
Thus bandwidth calculation and allocation can be done in a distributed manner

19. Multi-path QoS Routing
The algorithms searches for multiple paths between the source and the destination that collectively satisfies the QoS requirements
Suitable for ad hoc networks with limited bandwidth
A ticket based probing scheme is adopted for the path searching process

20. Transport Layer Issues for QoS Provisioning
TCP performs poorly in terms of end-to-end throughput in MANETs
The assumption used in Internet that packet losses are due to congestion is not valid in MANET environments
TCP performance improvement in wireless networks:
Local retransmissions
Split-TCP connections
Forward error corrections (FEC)
Explicit feedback mechanisms to distinguish between losses due to errors and congestion is necessary for QoS provisioning in MANETs
Efficient techniques for resource management is necessary for QoS provisioning

21. Application Layer Issues
Application level QoS adaptation belong to adaptive strategies that play a vital role in supporting QoS
Flexible user interfaces, dynamic QoS ranges, adaptive compression algorithms, joint source-channel coding, joint source-network coding schemes
Adaptive real-time audio/video streaming support can be provided by enhancing:
Compression algorithms, layered encoding, rate shaping, adaptive error control, and bandwidth smoothing

22. Inter-Layer Design Approaches
Efficient intercommunication protocols need to conserve scarce resources – something difficult to achieve following the strict separation of the protocol layer functionalities
Inter-layer or cross-layer issues needs to be examined
Examples: INSIGNIA and iMAQ

23. INSIGNIA
Goal: To support adaptive services which can provide base QoS assurances to real-time voice and video flows and data, allowing for enhanced level of service to be delivered when resources become available
Designed to adapt user sessions to the available level of service without explicit signaling between source-destination pairs
QoS functionality is decoupled from the routing protocol
INSIGNIA uses in-band signaling approach to restore the flow-state in response to topology changes
Uses the concept of “soft connection”

24. INSIGNIA Framework

25. iMAQ
Integrated Mobile Ad Hoc QoS (iMAQ) is a cross-layer architecture to support the transmission of multimedia data over a MANET
iMAQ framework model
Uses a predictive location-based QoS routing protocol
The middleware predicts the location and the group partitioning based on the moving pattern
Middleware may renegotiate QoS with applications when the resource availability degrades
Data is replicated when a network partition is predicted