QoS Routing in Ad Hoc Networks --Literature Survey Presented by: Li Cheng Supervisor: Prof. Gregor v. Bochmann Outline • • • • QoS routing overview: targets and challenges Classification of QoS routing protocols Typical QoS routing protocols Conclusion and Open Issues Video frame without QoS Support Video frame with QoS Support Li Cheng, ELG5125 Features of MANET • Mobile Ad-hoc Network • Definition: a self-configuring network of mobile routers (and associated hosts) connected by wireless links—the union of which form an arbitrary topology (www.wikipedia.org) • Features – – – – – Dynamic and frequently changed topology Self-organizing Nodes behaving as routers Minimal configuration and quick deployment Limited resources Li Cheng, ELG5125 Ad Hoc vs. Cellular Networks • Multi-hop route vs. One-hop route – In an Ad Hoc network, every nodes has to behave as a router • Self-administration vs. Centralized Administration – Ad hoc networks are self-creating, self-organizing, and self-administering PSTN OMC AC BSC GMSC MSC BSC HLR VLR BSC MS BTS BTS MS BTS Ad Hoc wireless network Cellular wireless network Li Cheng, ELG5125 Target of QoS Routing • To find a feasible path between source and destination, which – satisfies the QoS requirements for each admitted connection and – Optimizes the use of network resources <5,4> <4,5> A C B <4,2> <2,4> <3,3> D E <5,3> <3,4> <2,2> Tuple: <BW,D> F <4,4> G Li Cheng, ELG5125 QoS requirement: BW≥4 Shortest path QoS Satisfying path Challenges of QoS Routing in Ad Hoc Networks • • • • • • • • • • • Dynamic varying network topology Imprecise state information Scare resources Absence of communication infrastructure Lack of centralized control Power limitations Heterogeneous nodes and networks Error-prone shared radio channel Hidden terminal problem Insecure medium Other layers Li Cheng, ELG5125 Criteria of QoS Routing Classification • Routing information update mechanism – Proactive/table-driven: QOLSR, EAR – Reactive/On-demand: QoSAODV, PLBQR, TBP – Hybrid: CEDAR • Use of information for routing – Information of past history: QOLSR, QoSAODV, TBP – Prediction: PLBQR • State maintenance – Local: PLBQR, CEDAR – Global: TDMA_AODV, TBP • Routing topology – Flat: QOLSR, QoSAODV, PLBQR, TBP – Hierarchical: CEDAR • Interaction with MAC layers – Independent: PLBQR, QoSAODV, TBP – Dependent: CEDAR, PAR • Number of Path Discovered – Single path: QoSAODV, CEDAR, PLBQR – Multiple paths: TDMA_AODV, TBP • Utilization of Specific Resources – Power aware routing: PAR, EAR – Geographical information assisted routing: PLBQR Li Cheng, ELG5125 Typical Routing Mechanism • • • • • • Proactive routing: QOLSR Reactive routing: QoSAODV Ticket-based Routing: TBP Hierarchical Routing: CEDAR Predictive & Location-based routing: PLQBR Power aware routing Li Cheng, ELG5125 Proactive QoS Routing: QOLSR • Optimized Link State Routing[RFC3626] • Aiming at large and dense MANETs with lower mobility • Only selected nodes as multi-point relays (MPRs) forwards broadcasting messages to reduce overhead of flooding • MPR nodes periodically broadcast its selector list • QoS extensions – QOLSR[IETF Draft]: Hello messages and routing tables are extended with parameters of maximum delay and minimum bandwidth, and maybe more QoS parameters • Advantage: ease of integration in Internet infrastructure • Disadvantages: Overhead to keep tables up to date Black nodes: MPRs Li Cheng, ELG5125 Reactive QoS Routing: QoS Enabled AODV • AODV: Ad-hoc On-demand Distance Vector routing[RFC3561] • Best effort routing protocol • On need of a route, source node broadcasts route request(RREQ) packet • Destination, or an intermediate node with valid route to destination, responses with a route reply(RREP) packet. • QoS extensions[IETF Draft]: maximum delay and minimum bandwidth are appended in RREQ, RREP and routing table entry • Disadvantages – No resource reservation, which unable to guarantee QoS • Improved with bandwidth reservation: TDMA_AODV[7] – Traversal time is only part of delay RREQ1 (delay=100) Source Node A RREQ2 (delay=20) Rejected! RREQ1 (delay=70) Node B Traversal_time=30 Delay(B->D)=80 RREP1 (delay=80) RREQ1 (delay=20) Dest. Node D Node C Traversal_time=50 Delay(C->D)=50 RREP1 (delay=50) RREP1 (delay=0) QAODV example: Delay Requirement Li Cheng, ELG5125 Ticket-based Probing[5]: Features • Objective: To find delay/bandwidth-constrained least-cost paths • Source-initiated path discovery, with limited tickets in probe packets to decrease overhead • Based on imprecise end-to-end state information • QoS metrics: Delay and bandwidth • Redundancy routes for fault tolerance during path break • Destination initiated Resource Reservation p1(1) B p1(1) A C p2(2) p3(1) D p4(1) p4(1) E Li Cheng, ELG5125 Tickets-relative Issues • Colored tickets: yellow ones for smallest delay paths, green ones for least cost paths • For source node, how many tickets shall be issued? – more tickets are issued for the connections with tighter or higher requirements • For intermediate nodes, how to distribute and forward tickets? – the link with less delay or cost gets more tickets • How to dynamically maintain the multiple paths? – the techniques of re-routing, path redundancy, and path repairing are used Li Cheng, ELG5125 Disadvantages and Enhancement of TBP • Enhanced TBP Algorithm[13] – Color-based ticket Distribution – Ticket optimization using historical probing results Ticket blocking Color-based ticket distribution • Disadvantages – Based on assumption of relatively stable topologies – Global state information maintenance with distance vector protocol incurs huge control overhead – Queuing delay and processing delay of nodes are not taken into consideration Li Cheng, ELG5125 Hierarchical Routing: CEDAR[6] • Core Extraction Distributed Ad Hoc Routing • Oriented to small and middle size networks • Core extraction: A set of nodes is distributivedly and dynamically selected to form the core, which maintains local topology and performs route calculations • Link state propagation: propagating bandwidth availability information of stable high bandwidth links to all core nodes, while information of dynamic links or low bandwidth is kept local • QoS Route Computation: – A core path is established first from dominator (neighboring core node) of source to dominator of destination – Using up-to-date local topology, dominator of source finds a path satisfying the requested QoS from source to furthest possible core node – This furthest core node then becomes the source of next iteration. – The above process repeats until destination is reached or the computation fails to find a feasible path. Li Cheng, ELG5125 CEDAR: routing example G H A B C E Core Node D F S Links that node B aware of K J Node S informs dominator B G H A B C E D F S J Links that node E aware of Partial Route constructed by B K G Disadvantages of CEDAR: ― Sub-optimal route ― Core nodes being bottleneck H A B C E D F S J K Complete, with last 2 nodes determined by E Li Cheng, ELG5125 Predictive Location-based QoS Routing: PLBQR[8] • Motivation: to predict a future physical location based on previous location updates, which in turn to predict future routes • Update protocol: each node broadcasts its geographical update and resource information periodically and in case of considerable Predicted location change • Location and delay prediction: (xp, yp) at tp v(tp-t2) – Using similarity of triangles and (x2, y2) at t2 Pythagoras’ theorem, Direction of motion (xp,yp) can be calculated (x1, y1) at t1 – End-to-end delay from S to D is predicted to be same as delay of latest update from D to S • QoS routing – – – – Neighbor discovery with location-delay prediction Depth-first search to find candidate routes satisfied QoS requirements Geographically shortest route is chosen Route is contained in data packets sent by source • Disadvantages – No resource reservation – Inaccuracy in delay prediction Li Cheng, ELG5125 Power-aware QoS Routing • Objective: – to evenly distribute power consumption of each node – to minimize overall transmission power for each connection – to maximize the lifetime of all nodes • Power-Aware Routing[9]: using power-aware metrics in shortestcost routing – Minimize cost per packet, with cost as functions of remaining battery power – Minimize max node cost of the path to delay node failure • Maximum battery life routing[10]: Conditional Max-Min Battery Capacity Routing (CMMBCR) – To choose shortest path if nodes in possible routes have sufficient battery – Avoiding routes going though nodes whose battery capacity is below threshold • Energy Aware Routing[11]: selecting path according to its probability, which is inversely proportional to energy consumption, using sub-optimal paths to increase network survivability Li Cheng, ELG5125 Conclusion • QoS routing is key issue in provision of QoS in Ad Hoc networks • Number of QoS routing approaches have been proposed in literature, focusing on different QoS metrics • No particular protocol provides overall solution • Some Open Issues – – – – – – – QoS metric selection and cost function design Multi-class traffic Scheduling mechanism at source Packet prioritization for control messages QoS routing that allows preemption Integration/coordination with MAC layer Heterogeneous networks Li Cheng, ELG5125 Primary References [1] T.Clausen, P.Jacquet, Optimized Link State Routing Protocol(OLSR), IETF RFC3626, Oct.2993. [2] H.Badis, K.Agha, Quality of Service for Ad hoc Optimized Link State Routing Protocol (QOLSR), IETF Draft, Oct.2005 [3] C.Perkins, E. Royer and S. Das, Ad hoc On-Demand Distance Vector (AODV) Routing, IETF RFC3561, Oct.2993. [4] C.Perkins, E. Royer and S. Das, Quality of Service for Ad hoc On-Demand Distance Vector Routing, IETF Draft, Jul.2000. [5] S.Chen,K.Nahrstedt, Distributed Quality-of-Service Routing in Ad Hoc Network, IEEE Journal on Selected Areas in Commun, Aug 1999. [6] R.Sivakumar, P.Sinda and V. Bharghavan, CEDAR: A Core-Extraction Distributed Ad Hoc Routing Algorithm, IEEE Journal on Selected Areas in Commun, Aug 1999. [7] C.Zhu, M.Corson, QoS routing for mobile ad hoc networks, IEEE Infocom 2002. [8] S.Shah, K.Nahrstedt, Predictive Location-Based QoS Routing in Ad Hoc Networks, IEEE ICC 2002. [9] S. Singh, M.Woo and C.Raghavendra, Power-aware Routing in Mobile Ad Hoc Networks, MOBICOM’98. [10] C. Toh, Maximum Battery Life Routing to Support Ubiquitous Mobile Computing in Wireless Ad Hoc Networks, IEEE commun, Magazine, Jun 2001. [11] R Shah, J.Rabaey, Energy Aware Routing for Low Energy Ad Hoc Sensor Networks, IEEE WCNC 2002. Li Cheng, ELG5125 Secondary References [12] S.Chen,K.Nahrstedt, Distributed QoS Routing with Imprecise State Information, IEEE ICCCN’98. [13] L.Xiao,J.Wang and K.Nahrstedt, The Enhanced Ticket-based Routing Algorithm, IEEE ICC, 2002 [14] C.Murthy, B.Manoj, Ad Hoc Wireless Networks, Pentice Hall, 2004 [15] M.Ilyas, I.Mahgoub, Mobile Computing Handbook, Auerbach Publications, 2005 [16] S.Chakrabarti, A.Mishra, QoS Issues in Ad Hoc Wireless Networks, IEEE Commun. Magzine, Feb. 2001 Li Cheng, ELG5125 Any Questions? Li Cheng, ELG5125