1	Routing and QoS in MANETs Thomas Kunz Systems and Computer Engineering Carleton University http://kunz-pc.sce.carleton.ca/ tkunz@sce.carleton.ca
2	Mobile Ad Hoc Networks Infrastructure-less, may need to traverse multiple wireless links to reach a destination
3	Mobile Ad Hoc Networks Mobility causes route changes
4	Why Ad Hoc Networks ? Ease of deployment Speed of deployment Decreased dependence on infrastructure
5	Many Applications Personal area networking cell phone, laptop, ear phone, wrist watch Military environments soldiers, tanks, planes Civilian environments taxi cab network meeting rooms sports stadiums boats, small aircraft Emergency operations search-and-rescue policing and fire fighting
6	MANET Research Problems Wireless communication: huge field, but not mine J (I.e.: IEEE 802.11/15/16 standards, CDMA, coding and modulation) Many open research problems for TCP/IP networks, multiple IETF working groups exploring part of the problem space Dynamic topology: routing (MANET) and management (ZEROCONF) Mobility management (in particular when MANET is connected through gateway to fixed Internet): Mobile IP, SEAMOBY Transport layer: flow control for streaming services: DCCP Own research interest: routing (energy efficiency, capacity maximizing, multicast routing), QoS support, interlayer design
7	Routing: Open Problems Mobile Ad Hoc Network Routing Protocols: Limited radio transmission range and node mobility Many Routing Protocols (literally 100s): On-demand: AODV, DSR Pro-Active: OLSR Hybrid: ZRP Problems: Most existing performance comparisons based on ideal propagation model (free space model, two-ray ground model) Our simulation results with shadowing model show severe performance degradation
8	Propagation Models Question: does physical layer impact performance of routing protocols and if so, what to do about it? Ideal Models Free space model and two-ray ground reflection model Shadowing Model b: loss exponent, corresponding to mean transmission range s:shadowing deviation
9	Impact on Routing Protocols Signal strength fluctuates in shadowing model Ideal model (left) and shadowing model (right) over the same distance between two nodes Fluctuation at least 2 orders of magnitude
10	Impact on Routing Protocols Signal strength fluctuation causes active links to “break” Simulations of AODV & DSR show that Packet Delivery Ratio (PDR) decreases significantly Broken links introduce more Route Discovery processes Cause more control messages and longer packet delay
11	Simulation Results Performance Comparisons with different b values
12	QoS Routing: Find Routes satisfying QoS constraints Link state metrics should be available and manageable Link quality changes quickly and continuously due to node movement and surrounding changes Computational cost and protocol overhead affect the performance of the QoS routing protocol Protocol performance evaluation is complex
13	Proactive QoS Routing Advantages suitable for the unpredictable nature of Ad-Hoc networks suitable for the requirement of quick reaction to QoS demands makes call admission control possible avoids the waste of network resources Disadvantages introduces additional protocol overhead trade-off between the QoS performance and traditional protocol performance
14	Description of OLSR Selects MPR to cover 2-hop neighbors Exchanges neighbor/MPR information in Hello message Generates and relays TC message to broadcast topology information Reduces control overhead by limiting MPR set In the graph, B selects C as MPR
15	QoS Versions of OLSR OLSR protocol does not guarantee to find the best bandwidth route Three heuristics are proposed to enhance OLSR in bandwidth aspect The heuristics select good bandwidth neighbor as MPR Based on evaluation in static network scenarios, heuristic 2 is chosen: best-bandwidth neighbours are selected as MPRs until 2-hop neighbourhood is covered In the previous network topology, B selects A,F as MPRs
16	Analysis of Results Outperforms the original OLSR protocol in bandwidth aspect (i.e., finds better bandwidth routes) More MPRs are selected; more TC messages are generated and relayed The additional control messages increase the MANET network load The overlap of 2-hop neighbors covered by MPRs causes TC collision
17	Next Steps: Build Wireless MESH with QoS Support (joint project with NCIT and CRC) Mesh nodes: special hardware based on Intel IXP 425 NPU (ZAP nodes) Wireless Links: 802.11a/b/g, future support for 802.16a Software: Routing between ZAP nodes based on OLSR (CRC code) My research group: QoS support at MAC, routing, and end-to-end layer Related project (i.e., shared platform) at U of O with Prof. Makrakis
18
19