Slide 1

Ad Hoc Mobility Management with Uniform Quorum Systems

Slide 2

UQS Distributed mobility-management scheme For ad hoc Size of quorum intersection is a design parameter that can be tuned to adapt to the traffic and mobility patterns of the network nodes Tradeoff system reliability and cost of location update (more intersection more quorum size then increase traffic of location update) One possible solution partitioning of the network

Slide 3

Conventional HLR Conventional scheme HLR fixed association (node-HLR) For specific node it has specific HLR UQS dynamic association (node-HLR)

Slide 4

Virtual backbone Location databases which form a virtual backbone Routing does not necessarily go through the databases The node location provided by the databases can provide vital information to the routing protocol Define the location of a mobile host as the ID number of its nearest location database

Slide 5

Virtual backbone Need to maintain connectivity Due to backbone nodes are few compared with the whole network the extra overhead causes by routing is less

Slide 6

node-DB association Nodes containing the location databases can dynamically detach and reattach to the network at any time due to mobiles movements or changes in the communication environment

Slide 7

node-DB association Fail whenever the node containing the database detaches from the network for a long time Fixed association between nodes and databases (node-DB association)

Slide 8

node-DB association Avoid fail problem multiple HLRs (but still has fixed association problem) Solution virtual HLR or adaptive HLR (Randomly choose a node as a new HLR and the database name of this HLR is still the same)

Slide 9

Dynamics of UQS UQS dynamic association (node-HLR) between any particular quorum to any mobile host May choose the quorum contains the nearest database and this kind of quorum may be several ones and randomly choose one of them as HLR

Slide 10

Dynamics of UQS Difference with multiple HLR If all multiple HLRs of the node are fail then fail to the node As long as at least one complete quorum is survival then the system is not fail Each quorum may (can) be the HLR of the node

Slide 11

Dynamics of UQS Because the intersection property of quorum UQS is dynamic (no fixed association between any particular quorum to any mobile host) (node- HLR)

Slide 12

Others of quorum Quorum is more flexible than adaptive HLR for example fixed node-HLR association others need to know your HLR Quorum has high reliability because at least one quorum survives Other advantage location and service provider independent numbering load balancing

Slide 13

Others of quorum BIB (Balanced Incomplete Block) design problem is the same as UQS construction Cost cost to accessing databases call loss due to database failure But large intersection size require larger size of quorums, which increase the cost of location updates and location queries

Slide 14

Fail & Inaccessible When a backbone node detached from the network a nearby non-backbone node is recruited to take its place in the virtual backbone When detach exceeds the threshold (time) then replacement But the original database in the node is lost we call this database fail Inaccessible detach doesnt not exceed the threshold

Slide 15

Fail & Inaccessible Need a mechanism to check accessibility of nodes and the frequency of checking is much higher than that of location update and query (and because the number of virtual nodes is few the cost is low) Number of intersections will affect backup and reliability Fail and inaccessible should be solved

Slide 16

Update location Situation of update location (1) When using the quorum (read) update the quorum (2) change of quorum (due to change of location or other reasons) (3) periodic

Slide 17

Call loss Call loss (1)database fail (may have erroneous information) (2)database inaccessible (low probability due to the time of inaccessible is low) The probability of call loss depends on the stability of network the frequency of update (but more overhead)

Slide 18

Construction of UQS Five parameters for UQS construction N number of database Q number of quorum K size of quorum Mi number of quorum contain database i R number of intersection Represented by a N*Q matrix entry will be 0 or 1 On purpose of uniform each (database or quorum) R K M is the same

Slide 19

Construction of UQS Some properties of UQS Parameter is dependent For a given set of parameters its UQS may not exist UQS is the same as BIB so solution to BIB is solution to UQS

Slide 20

Construction of UQS Due to no sufficient condition of BIB there is no way to decide whether UQS exists or not And there is no formula can represent the relation among N K R so it can only be studied some UQS parameter sets under certain situation

Slide 21

Construction of UQS Five series(skip now) Given N show the allowable variation of r in Fig3 (the five series) But the five series cant contain all situations In fact the choice of N will be affected by routing and load Even given R the relation of N-K is not one-to- one

Slide 22

Construction of UQS

Slide 23