Routing in Mobile Networks Professor Ching-Chi Hsu Part I. Mobile IP on Network Layer Part II. Routing in Mobile/Wireless Ad-hoc Networks 0 Mobile IP on Network Layer DNS based Name to Address resolution Network Layer Mobility Problem Illustration of Terms in Mobile IP Two Tier Addressing Mobile IP Architecture Components Address Translation Mechanisms Mobile IP Proposals 1 DNS based Name to Address resolution 2 Network Layer Mobility Problem Directory Service View DNS didn’t handle dynamic updates. DNS design attempts to optimize the access cost, and not the update cost. There is no call back mechanism generally available from servers to clients in case the cached entries of the DNS clients become invalid. 3 Network Layer Mobility Problem Internet View TCP connection: <source IP address, source TCP port, destination IP address, destination TCP port> If the mobile host acquires a new IP address, all TCP connections involving the mobile host will be broken. If the mobile host retains its address, then the routing system cannot forward packets to its new locations. 4 Illustration of Terms in Mobile IP 5 Two Tier Addressing 6 Mobile IP Architecture Components 7 Address Translation Mechanisms Encapsulation 8 Address Translation Mechanisms Loose Source Routing 9 Mobile IP Proposals Columbia Scheme 10 Mobile IP Proposals Sony Scheme 11 Mobile IP Proposals LSR Scheme 12 Mobile IP Proposals IETF Scheme 13 Routing in Mobile/Wireless Ad-Hoc Networks Introduction Definitions of Dynamic Groups Routing Dynamic Groups Routing Algorithms Simulation and Analysis Conclusion Future Works 14 Introduction Ad-Hoc Networks Model Previous Works on Routing in Ad-Hoc Networks Motivations of Dynamic Groups Routing 15 Ad-Hoc Networks Model Ad-hoc networks Wireless communication Without fixed network interaction and centralized administration. Multi-hop routing style with no cheating of the forwarding nodes mobile host routing 16 Ad-Hoc Networks Model Why ad-hoc networks ? When the access base station are not set up due to low cost effect, poor performance or low usage Ad-hoc networks may be the extension of base stations Base station Bridges Ad-hoc networks Applications of ad-hoc networks An outdoor conference Emergency situations of natural disasters Military deploys in battlefield 17 Previous Works on Routing in Ad-Hoc Networks On-demand dynamic source routing (D.B.Johnson et al.) Cluster-based routing (P. Krishna et al.) Zone routing (Z.J. Haas et al.) Minimum connected dominating set routing (B. Das et al.) 18 On-demand dynamic source routing Loosely source routing while route requested Each node may overhear the routing information from the neighbor nodes. Example: source routing and overhearing source destination Route request message Route reply message Mobile node Overhearing range 19 Cluster-based routing The clustering method depends on the k-hop mutually reachable relation between any two nodes in a cluster. A two-level network graph - cluster-level + node-level Example: 1-cluster(fully connected sub-graphs as clusters) Mobile node Cluster 20 Zone routing Zone - the nodes within the defined radius Radius - hop-count reachable from the central node Each node maintain the topology in its zone Example: zone with radius = 2 Mobile node Zone 21 Minimum connected dominating set routing Minimum connected dominating set (MCDS) - construct a so-called virtual backbone in an ad-hoc network. Routing through the virtual backbone by dominators. Example: virtual backbone construction MCDS edge Non-MCDS edge Mobile node dominators MCDS(virtual backbone) 22 Comparison Routing methods Infrastructure Advantages Disadvantages On-demand Routing Only sense the neighbors Less overhead to Frequent routing maintain topology make it mass route discovery overhead Zone Routing (cluster-based) Uniform cluster with the node as the center Easy to maintain the topology in zones MCDS Routing MCDS virtual backbone as routing trunk Simple to search routes by virtual backbone 23 Frequent moving causing heavy topology update Network grown up make the virtual backbone a heavy burden Motivations of Dynamic Groups Routing Ad-hoc network topology maintenance Range Consideration The whole network - inefficient and infeasible. Only neighbors - inefficient while frequent route requesting Structure Consideration Relative conectivity Relative position To propose an easy topology maintenance and good traffic isolation ad-hoc network routing system- Dynamic Groups Routing 24 Definitions of Dynamic Groups Routing Illustration of an ad-hoc network construction Definitions Dominating value definition Routing groups definitions Definitions of attachable sets, bridge and peripheral clusters 25 Illustration of an ad-hoc network construction A B C D H N E I K J L F G P M O Q R Positive cluster Non-positive cluster Routing group 26 Dominating value definition Definition 1: DV(p) is the dominating value of a node p in an ad-hoc network. It is calculated as the following before the construction proceeds. Initial DV(p) := 0 For each q which q in neighborhood of p If deg(p) > deg(q) then DV(p) := DV(p) +1 else if deg(p) < deg(q) DV(p) := DV(p) – 1 Theorem 1: In the initial constructed bi-directional connected network, the summation of all dominating values of nodes is zero. 27 Routing groups definitions Definition 2: A node p is positive while DV(p) > 0, otherwise non-positive. Definition 3: In a constructed ad-hoc network, the cluster formed by all connected positive nodes is called positive cluster (P-cluster). The cluster formed by all connected nonpositive nodes is called non-positive cluster (N-cluster). Definition 4: A routing group (RG) is formed by one Pcluster with its adjacent N-clusters 28 Definitions of attachable sets, bridge and peripheral clusters Definition 4: An attachable set is a set of all the positive nodes in the same P-cluster with direct bi-directional links to the same adjacent N-cluster Definition 5: A bridge cluster is the N-cluster belonged to more than one RG. On the other hand, the N-cluster belonged to one RG is called peripheral cluster. 29 Dynamic Groups Routing Algorithms Construction algorithm Route discovery strategy Route maintenance Connection recovery 30 Message propagating illustration in construction <2,-2> Nlist(A),DV(A) <3,0> <3,-2> B M A <3,0> D H <5,+2> <2,-2> Q <2,0> R <2,-2> <4,4> <5,+2> C P N O <3,-2> <5,+2> <5,+2> E I K <2,-1> F <2,-1> G J <2,-1> <2,-1> 31 L <2,0> Note: <degree, DV> Construction algorithm {For each mobile node p in the ad-hoc network} Broadcast message deg(p) to all neighbors Receive message deg(q) from each neighbor q Calculate DV(p) as definition 1 Broadcast message {List(N(p)) ,DV(p)} to each neighbor q Receive message {List(N(r)) ,DV(r)} from each neighbor q For each message {List(N(r)) ,DV(r)} Case 1. DV(p) > 0 & DV(q) > 0 Propagate it to all positive neighbors excluding q Case 2. DV(p) > 0 & DV(q) <= 0 Propagate it to all positive neighbors Case 3. DV(p) <= 0 & DV(q) <= 0 Propagate to all neighbors excluding q Case 4. DV(p) <= 0 & DV(q) > 0 Ignore the message 32 Route discovery illustration Intra-cluster Intra-group Inter-group Intra-group Inter-group Intra-cluster Mobile node S RG P-cluster N-cluster 33 Route discovery strategy For a route request from the source node s to the destination node d if (the node s is non-positive) then it checks the local topology information of its N-cluster if (the destination node d is found) then it routes from the node s to the node d directly else it multicasts intra-group route request message to all the nodes in its attachable sets and waits for reply message if (all the reply message are failure) then it issues inter-group route request message to all the nodes in its attachable sets else it routes by the reply information from one of the nodes in its attachable sets else it checks the local topology information of its routing group if (the destination node d is found) then it routes from the node s to the node d directly else it multicasts inter-group route request message to all the nodes in the attachable sets of all its bridge clusters,which also belong to other routing groups 34 Route maintenance(link variations) 6 kinds of reactions to the variations Reacting/Interacting node Interacting/Reacting node Link Hm Hn Connected/Disconnected N/P 2 N/P * 2 * 35 2 =8 Route maintenance(link variations) Hm P Hn P P N N N P P N P P N N N P N Link(Hm,Hn) Action Description Disconnect I Hm propagates "Delete link(Hm,Hn)" to the hosts in the same P-cluster. If the P-cluster isn’t connected, the RG is splited. Disconnect II Hm propagates "Delete link(Hm,Hn)" to the hosts in the same P-cluster. Hm is detached the N-cluster of Hn from Hn . Disconnect III Hm propagates "Delete link (Hm,Hn)" to the hosts in the same N-cluster or in its ASET Connect IV If Hn and Hm are the same P-cluster, Hm propagates "Insert link(Hm,Hn)" to the hosts in the same P-cluster. Otherwise Hm propagates "Insert RG topology of Hm" to Hn and waits for the RG topology of Hn. Therefore, the action joins two RGs. Connect V Hm propagates "Insert link(Hm,Hn)" to the hosts in the same P-cluster. If Hn and Hm are not the same RG, Hm waits for the N-cluser+ASET topology of Hn. Connect VI Hm propagates "Insert link (Hm,Hn)" to the hosts in the same N-cluser or in its ASET. If Hn and Hm are not the same RG, Hm propagates "Insert N-cluser+ASET topology of Hm" to Hn. 36 Route maintenance (connectivity dominating variations) -1 DV(E): +2 G A B D E F H I C 37 Route maintenance (connectivity dominating variations) The connectivity dominating variation of Hm caused by the link variation of Hn Initial DV(Hm) Deg(Hm)? deg(Hn) > P = P < P > N = N < N > P = P < P > N = N < N LinkHmHn variation Connected→disconnected Connected→disconnected Connected→disconnected Connected→disconnected Connected→disconnected Connected→disconnected Disconnected→connected Disconnected→connected Disconnected→connected Disconnected→connected Disconnected→connected Disconnected→connected Note: P: positive ; N: non-positive; →: link state transition from left to right 38 Updated DV(Hm) P or N P P N N N or P P P P or N N or P N N Route maintenance (connectivity dominating variations) The connectivity dominating variation of Hm caused by the conectivity dominating variation of Hn Initial V(Hm) Deg(Hm)?deg(Hn) P > P = P < N > N = N < P > P = P < N > N = N < DV(Hn) variation Decrement Decrement Decrement Decrement Decrement Decrement Increment Increment Increment Increment Increment Increment Note: P: positive ; N: non-positive 39 Updated DV(Hm) P P P or N N N or P N or P P or N P or N P N or P N N Connection Recovery Group-level backtracking to the last one New route N-cluster N-cluster N-cluster P-cluster RG1 P-cluster RG2 40 Connection Recovery Assume original discovered route from s to d. (s(a0)a1… an d(an+1)) The corresponding attachable sets are A1,A2,….,An For node ai Search the topology information in the routing group of ai If the destination node d is found Routing is successful Otherwise Checking the connection status of ai+1 Case “Normal” Routing performs on the next node ai+1. Case “Broken” (Recovery process performs on node ai) Finding the other attachable sets from the routing group of ai. If there are other attachable sets excluding the routing group of ai-1 and ai+1 Rerouting starts from these attachable sets. Otherwise Routing backtracks to ai-1 to find the other attachable sets excluding the routing group of ai. Rediscovery process performs on ai-1. Case “Merged” Forwarding the routing process to ai+1. Case “Attached” Forwarding the routing process to the attachable set excluding the routing group of ai. 41 Simulations and Analysis Simulation model Simulation results 42 Simulation model N mobile nodes move randomly in a square area 43 Simulation results Connectionless Variant number of mobile hosts Variant moving frequency of mobile hosts Variant route query frequency of mobile hosts Connection-oriented 32 mobile hosts in 800m*800m area 72 mobile hosts in 1200m*1200m area 44 Total delay in route maintenance phase by increasing number of mobile hosts Message Trip for Maintenance 70000 60000 DG Routing(mtc) Zone Routing(mtc) MCDS Routing(mtc) Total hop count 50000 40000 30000 20000 10000 0 10 20 30 40 50 60 Number of mobiles 45 70 80 90 100 Control traffic in route maintenance phase by increasing number of mobile hosts Message Propagation of Maintenance 3000000 Number of message packets 2500000 DG Routing(mmc) Zone Routing(mmc) MCDS Routing(mmc) 2000000 1500000 1000000 500000 0 10 20 30 40 50 60 Number of mobiles 46 70 80 90 100 Control traffic in route discovery phase by increasing number of mobile hosts Message Propagation of Route Discovery 20000000 Number of propagated message packets 18000000 DG Routing(rmc) Zone Routing(rmc) MCDS Routing(rmc) Flood Routing(rmc) 16000000 14000000 12000000 10000000 8000000 6000000 4000000 2000000 0 10 20 30 40 50 60 Number of mobiles 47 70 80 90 100 Total control traffic by increasing number of mobile hosts Message Propagation of Route Maintenance & Discovery 20000000 Number of propagated message packets 18000000 DG Routing(tmc) Zone Routing(tmc) MCDS Routing(tmc) Flood Routing(tmc) 16000000 14000000 12000000 10000000 8000000 6000000 4000000 2000000 0 10 20 30 40 50 60 Number of mobiles 48 70 80 90 100 Total delay in route maintenance phase by increasing moving frequency Message Trip for Maintenance 35000 30000 Total hop count 25000 20000 15000 10000 DG Routing(mtc) Zone Routing(mtc) MCDS Routing(mtc) 5000 0 10 20 30 40 50 60 70 Frequency of movement(%) 49 80 90 100 Control traffic in route maintenance phase by increasing moving frequency Message Propagation of Maintenance 600000 DG Routing(mmc) Zone Routing(mmc) MCDS Routing(mmc) Number of message packets 500000 400000 300000 200000 100000 0 10 20 30 40 50 60 70 Frequency of movement(%) 50 80 90 100 Control traffic in route discovery phase by increasing moving frequency Message Propagation of Route Discovery Number of propagated message packets 3000000 2500000 2000000 DG Routing(rmc) Zone Routing(rmc) MCDS Routing(rmc) Flood Routing(rmc) 1500000 1000000 500000 0 10 20 30 40 50 60 70 Frequency of movement(%) 51 80 90 100 Total control traffic by increasing moving frequency Message Propagation of Route Maintenance & Discovery Number of propagated message packets 3000000 2500000 DG Routing(tmc) Zone Routing(tmc) MCDS Routing(tmc) Flood Routing(tmc) 2000000 1500000 1000000 500000 0 10 20 30 40 50 60 70 Frequency of movement(%) 52 80 90 100 Total delay in route maintenance phase by increasing routing frequency Message Trip for Maintenance 35000 30000 Total hop count 25000 20000 15000 10000 DG Routing(mtc) Zone Routing(mtc) MCDS Routing(mtc) 5000 0 10 20 30 40 50 60 70 Frequency of route request(%) 53 80 90 100 Control traffic in route maintenance phase by increasing routing frequency Message Propagation of Maintenance 400000 350000 Number of message packets 300000 250000 DG Routing(mmc) Zone Routing(mmc) MCDS Routing(mmc) 200000 150000 100000 50000 0 10 20 30 40 50 60 70 Frequency of route request(%) 54 80 90 100 Control traffic in route discovery phase by increasing routing frequency Message Propagation of Route Discovery Number of propagated message packets 60000000 DG Routing(rmc) Zone Routing(rmc) MCDS Routing(rmc) Flood Routing(rmc) 50000000 40000000 30000000 20000000 10000000 0 10 20 30 40 50 60 70 Frequency of route request(%) 55 80 90 100 Total control traffic by increasing routing frequency Message Propagation of Route Maintenance & Discovery 60000000 DG Routing(tmc) Zone Routing(tmc) MCDS Routing(tmc) Flood Routing(tmc) Number of propagated message packets 50000000 40000000 30000000 20000000 10000000 0 10 20 30 40 50 60 70 Frequency of route request(%) 56 80 90 100 The message complexity comparison by connection time duration with area = 800m*800m, number of mobile nodes = 32, moving frequency =50% Message Complexity 40000 DG without recovery DG with recovery Flooding routing Zone routing MCDS routing number of propagated message packets 35000 30000 25000 20000 15000 10000 5000 0 10 20 30 40 50 60 70 connection time duration(#clicks) 57 80 90 100 The message complexity comparison by mobile node moving frequency with area = 800m*800m, number of mobile nodes = 32, connection duration = 50 Message Complexity 20000 number of propagated message packets 18000 16000 14000 12000 DG without recovery DG with recovery Flooding routing Zone routing MCDS routing 10000 8000 6000 4000 2000 0 10 20 30 40 50 60 moving frequency % 58 70 80 90 100 The message complexity comparison by connection time duration with area = 1200m*1200m, number of mobile nodes = 72, moving frequency =50% Message Complexity number of propagated message packets 140000 DG without recovery DG with recovery Flooding routing Zone routing MCDS routing 120000 100000 80000 60000 40000 20000 0 10 20 30 40 50 60 70 connection time duration(#clicks) 59 80 90 100 The message complexity comparison by mobile node moving frequency with area = 1200m*1200m, number of mobile nodes = 72, connection duration = 50 Message Complexity 160000 DG without recovery DG with recovery Flooding routing Zone routing MCDS routing number of propagated message packets 140000 120000 100000 80000 60000 40000 20000 0 10 20 30 40 50 60 70 moving frequency % 60 80 90 100 Conclusion A two-type cluster scheme is proposed to form the routing groups structure, easily maintaining the routing environment in ad–hoc networks. It seems that this approach is adaptive to such domain for reducing the control message propagation in ad-hoc networks Upon the basic constructed model of ad-hoc networks, the route recovery method is developed for connection-oriented applications 61 Future Works Now a day many multicasting applications are applied in mobile/wireless ad-hoc networks. The next research subject is developing efficient multicast routing methods in ad-hoc networks based on the constructed dynamic groups model. 62