A Proposal for an
Ad-Hoc Network
 QoS Gateway

Agenda

	Introduction and Motivation
	Problem Definition & Major Challenges
	Design Fundamentals
	Proposed Model Architecture
	Proposed Model Performance and Behavioral Characteristics
	Summery of Simulation Results and Conclusion
	Publications Resulted from this Research

Motivation

	The increasing availability of WLAN.
	The enormous growth of cellular networks.
	The development of Community Access Networks.
	The growing popularity of P2P wireless applications.
	The increasing interest in ad-hoc and MESH networks.
	All of the above market growth fuel the need to develop a QoS access gateways that can provide better than BE services to roaming nodes with minimum configuration.

QoS Approaches from Layered Perspective

	MAC QoS
		DCF, EDCF, WLAN implementations
	QoS-aware routing
		AODV, DSR, OLSR,
	Inter-layer QoS models
		dQoS, FQMM, INSIGNIA, SWAN
	QoS-aware Applications
		Middleware, VoIP codecs
	Other approaches
		FPRP, RSVP-MP
	CROSS-DOMAIN view: IntServ over DiffServ

Challenges to QoS Access Gateway

	The merging of different QoS model employed on each side of the gateway.
	The lack of a policing authority in the ad-hoc domain leads to the need for extra policing on the gateway that may not be scalable.
	The difficulty in maintaining accounting and billing records for the mobile nodes due to the lack of a central authority.
	The difficulty in obtaining effective service provisioning leaves the gateway with a hard optimization decisions.
	Designing scalable architecture.

Proposed Model Cases

Design Fundamentals

	The Reuse of Common Gateway Design Principals
		Common Service Subset, Asymmetric Design
	Access Network as a Private Friendly DiffServ Domain
		E2E ARSVP reservation
	Reactive Aggregated Resource Allocation
		Services are requested  only when needed
	The Use of Limited Policing
		Compromise between scalability and policing
	Service Ladder Policy (Optional)

Gateway Architecture in Upstream Scenario

Proposed Model (Disabled) in DNST – Bandwidth View

Proposed Model (Enabled) in DNST – Bandwidth View

Why We Improve Both RT and BE-bandwidth

Proposed Model in DNST – Delay View

Conclusion on Proposed Model Behavior

	Enabling proposed model increases the amount of bandwidth assigned to RT-flows. In association, the BE-bandwidth experience limited change.
	Enabling proposed model decreases the average RT-delays by a large margin, while the average BE-delays experience limited change.
	Enabling proposed model results in smoother bandwidth and delay charts which facilitates smaller jitter.

Comments On Proposed Model Performance

	The overall gain when using the proposed model depends highly on the various components used in the model. The model has been tested over DiffServ, INSIGNIA, SWAN and ESWAN.
	The performance of the ad-hoc QoS implementation has high impact on the overall performance due to the limited resources available on ad-hoc domains.
	The use of reactive resource allocation avoids the trouble of provisioning services, but adds a little latency to service initiation.
	The proposed model shows significant enhancement to supported RT traffic BW & delay associated with limited impact on BE traffic when employing SWAN.

Publications Resulted from this Research

	Yasser L. Morgan and Thomas Kunz, “A Proposal for an Ad-hoc Network QoS Gateway”, to  appear in the Proceedings of the IEEE International Conference on Wireless and Mobile Computing, Networking and Communications WiMob-05, Montreal Canada, August 2005.
	Yasser L. Morgan and Thomas Kunz, “A Design Framework for Wireless MANET QoS Gateway”, to  appear in the Proceedings of the 6th ACIS International Conference on Software Engineering, Artificial Intelligence, Networking, and Parallel/Distributed Computing, pp. 31-37, Towson USA, May 2005.
	Yasser L. Morgan and Thomas Kunz, “Enhancing SWAN QoS Model By Adopting Destination-Based Regulation (ESWAN)”, in the Proceedings of the 2nd WiOpt-04 Conference for Modeling and Optimization in Mobile Ad-hoc and Wireless Networks, pp. 112-121, Cambridge UK, March 2004.
	Yasser L. Morgan and Thomas Kunz, "PYLON: An architectural framework for ad-hoc QoS interconnectivity with access domains", in the Proceedings of the 36th International Conference on System Sciences, Hawaii USA, January 2003.
	Yasser L. Morgan and Thomas Kunz, “An Architectural Framework for MANET QoS Interaction with Access Domains”, Proceedings of the 1st International Conference on Ad-Hoc and Wireless Networks, pp 33-47, Toronto Ontario-Canada, September 2002.

Discussion & Remarks