Mobility and Location Management

1
Mobility and Location Management

• Mobility Management Problems
• Handoff Detection and Operations
• Location-Dependent Queries
• Location Update Methods and Data Modeling
• Example Location Management in Cellular Network
• Paging and Paging Strategies
• Location Area Planning

2
Mobility Management Problems

• Mobile Computing Vs. Distributed Computing
• Connection (wireless) with mobility capability
• Handoff Vs. Roaming (not in connection)
• Handoff (in connection)
• Providing continuous connection to a mobile
  station while moving
• from a cell into another cell
• from a network into another network
• When a mobile user moves into the coverage area
  of another network (cell), the radio link to the
  old network is eventually disconnected (becoming
  weak)
• A new radio link to the new network should be
  established to maintain the connection quality

3
Handoff Operations

• Handoff operations (MT -gtBSs-gt MSC)

4
Mobility Management Problems

• Roaming
• Moving but not in connection
• location update and paging
• Keeping the current location of an object while
  it is moving
• Positioning techniques (I.e., GPS and cell ID
  updates)
• Location update (registration), when? under which
  conditions?
• Efficient link connection to a moving object
  (calling a mobile station)
• For example, in a PCS, if a mobile station wants
  to connect to another mobile station, how??
• Search for the called mobile station by paging
• What are the cells to be paged?
• A database maintained the previous reported
  locations of the mobile stations (location
  database)
• Location update cost proportional to the number
  of update generated
• Paging cost depends on the number of areas to be
  searched
• Location update cost Vs. paging cost
• Searching delay (paging delay)

5
Detection of Handoff

• Handoff operations are expensive and need to be
  minimized
• Selection and then establish the new link with
  the new base station
• A mobile user may be at a location that it may
  connect to several networks. Which network should
  it choose to connect to?
• Removal of the old connection
• What will be the problem if a handoff cannot be
  performed?
• I.e., the new base station does not have any free
  channel for connection (forced termination)
• To initiate a handoff, two issues must be
  considered
• Who initiates the handoff operation?
• How is the need for handoff detected?
• Handoff detection is based on link quality
  measurement
• It determines the needs for handoff and the new
  channel for transfer
• Link quality affected by many factors and always
  not stable diffusion, reflection and multi-path
  propagation (fading problems)
• Techniques such as frequency hopping are employed
  to make the link quality more stable (changing to
  another frequency when it is not stable)

6
Detection of Handoff

• Mobile-controlled handoff (MCHO)
• The mobile user continuously monitors the
  strength of the signals of the surrounding base
  stations and initiates the handoff process when
  some handoff criteria are met
• Network-controlled handoff (NCHO)
• The surrounding base stations (BSs) measure the
  signal from the mobile user, and the network
  initiates the handoff process when some handoff
  criteria are met
• Mobile-assisted handoff (MAHO)
• The network asks the mobile user to measure the
  signal from the surrounding BSs. The network
  makes the handoff decision based on the reports
  from the mobile station (MS)

7
(No Transcript)
8
Handoff Failure and Channel Assignment Schemes

• Handoff failure
• No channel is available on the selected base
  station
• Normally, the number of channels in a cell or the
  total bandwidth in a network is fixed (Vs.
  dynamic channel allocation scheme)
• Channel Assignment Strategies
• Minimize the probability of handoff failure due
  to no free channel (forced termination)
• Reservation Vs. no reservation (same as new call)
• Channel reservation may increase the number of
  call blocking (from new call connection)
• What is the number of channels to be reserved for
  handoff users?
• The number of channels handled by a base station
  is normally fixed
• Two types of requests for channels handoff and
  new call

9
Handoff Failure and Channel Assignment Schemes

• Queuing priority scheme handoff operation may be
  delayed due to overlapping service areas
  (degradation interval and handoff area)
• FIFO
• Measured-based priority scheme (MBPS)
• Priority defined based on power level received
  from the links
• Higher priority to the mobile station with a
  weaker link quality
• Channel Assignment in shared networks
• A connection may require more than one channel,
  i.e., in multimedia transmission
• How many number of channels to be assigned?
• If insufficient number of free channels, what to
  do?
• Look at the one with smaller channels requirement
• A guarantee in quality of services
• The number of channels required for a connection
  may be included in the initial call for
  connection

10
(No Transcript)
11
(No Transcript)
12
(No Transcript)
13
Importance of Mobility Management

• Location-dependent services
• Different services are provided at different
  areas to different users
• Different areas have different application
  characteristics and the users will have different
  types of requests (spatial properties)
• Sales news broadcasting in shopping malls
• Traffic and weather information at different
  areas (Central, Kowloon Tong, )
• Active detection of location information of
  clients
• Enter a shopping mall -gt your location is updated
  -gt rule-evaluation -gt location-dependent services
  (big sales information)
• Positioning techniques
• For cellular phone users using cell ID
• Global positioning system (GPS)
• Smart devices, i.e., RFID and active badge
  (location sensor system, http//www.ics.agh.edu.pl
  /ABng/)

14
Location-Dependent Applications

• Example applications
• Automobile navigation, traffic information,
  etc..
• Fleet tracking
• Trucking companies use LD services to minimize
  the cost and maximize productivity (courier
  services)
• Helpful for dispatching taxis
• Advertising and marketing
• Discount and promotional coupons
• Agriculture and environmental protection
• Identity the living habit of animals and the
  health conditions of the plant at different
  regions
• GPS map and development of harvest strategies
• Animal tracking
• Security and theft control
• Track stolen items
• Accessibility
• Lost people (old man)
• Many more

15
Location-Dependent Queries

• Queries processing for supporting various
  location dependent services LDQ and LDCQ
• Location-dependent queries (LDQ)
• The location information of a moving client is
  submitted together with the query
• The result to be returned depends on the location
  of the client
• I.e. What is the nearest Italian restaurant from
  my current position (in Kowloon Tong Vs.
  Tsimshatsui)
• Location-dependent continuous queries (LDCQ)
• The LDQ is submitted with a begin time and an end
  time (i.e., from now to 10 min later)
• Continuously monitor the database throughout the
  activation period (i.e. re-evaluation after each
  update)
• What is the nearest Italian restaurant from my
  current position from now on until 10 min later
  (re-evaluate after each movement)
• I.e. Navigation for the shortest path to the
  nearest hospital

16
Location-Dependent Queries

• LDCQ
• From a moving object
• Searching for a car park
• From a stationary object on moving objects (or
  other dynamic information)
• EQ1 A police officer may ask How many patrol
  cars are within 2 km of the airport?
• From a moving object on moving objects
• A patrol car is looking for other patrol cars

17
Centralized Vs. Distributed Management

• Centralized approach use a powerful server to
  manage the space status and connection
  information
• Distributed approach multiple devices (service
  providers) manage the information
• Comparisons
• Problems in distributed computing
• Perform operations at device level because of
  limited bandwidth
• Due to the dynamic properties of the smart space
  and objects, a lot of updates are needed to be
  generated
• A distributed approach can make the management of
  objects to be localized and adaptive to the
  changing systems status (in networking
  processing). But, the communication overhead
  could be very heavy
• A hierarchical approach multiple levels with
  different types of coordinators may be used

18
Location Management Architecture
19
Location Data and Representation

• Location databases
• A database maintains location information of all
  moving objects (i.e., coordinates, identify, ..)
• The objects periodically/conditionally (how?)
  generate location update to refresh their
  locations maintained in the location database (to
  minimize the number of updates)
• Location information
• Coordinates gt X 23.111, 104.202
• Vs Location area ID gt X in school
• Accuracy Vs cost in processing
• In reporting the query results, we usually not
  simply present the coordinates
• The accuracy is highly affected by the size of
  each location area and how they are divided

20
Location Update and Evaluation Problems

• Query result problem
• If the location of any objects accessed by a
  query changes, the result of the query may change
  too
• For example, EQ1 asking the number of cars within
  2km of the airport
• Ten cars satisfy EQ1 at time t1. At time t2
  (where t2 t1d), a car has moved and it is no
  longer within 2km of the airport
• Thus the result of EQ1 at time t2 has to be
  re-evaluated
• But, the location database may still keep its old
  location and no re-evaluation will be performed
• The client may still believe that the old results
  are still valid
• Evaluation problem
• Re-evaluate the query whenever an update has
  arrived? Heavy evaluation cost if the update rate
  is high but most of them are similar
• Periodic re-evaluation? What is the best period?
  Long Vs short
• Data transmission volume
• Proportional to the number of updates

21
(No Transcript)
22
Location Update Methods

• Time-based
• A periodic update scheme
• Generate an update every fixed time threshold T
• How to define T? Based on the number of updates
  that the system can support
• Distance-based
• If the difference between the current distance
  and the last update location is larger than D, an
  update is generated
• Hybrid (time-based distance-based)
• Either condition from TB and DB
• Speed-dead-reckoning
• An update is generated if the deviation of its
  current location is greater than the predicted
  location by a pre-defined distance threshold
• Location area
• The system is divided into inter-connected
  location areas
• An update is generated if a moving object enters
  into a new LA
• How to divide the areas? (Location area planning
  problem)
• How to organize the location areas to minimize
  the searching cost

23
Time-based Location Update Method
Uncertainty in location Vs. no. of updates
24
Distance-based Location Update Methods
Uncertainty in location Vs. no. of updates
No update because of disconnection?
25
Hybrid Method (time-based distance-based)
Reset the threshold whenever an update is received
26
Speed-dead-reckoning (SDR)
Number of updates depends on the accuracy in
prediction
27
Location Area (LA)
What will be the problem if an object is moving
in the boundary?
28
Modeling of Moving Objects

• Moving Object Spatio-Temporal (MOST)
• For location management and for location
  prediction
• To minimize the update cost (frequency)
• A predictive approach
• Attributes static dynamic
• Dynamic attribute A of a moving object
• A.value A.updatetime A.function
• At time A.updatetime, the value of A is A.value
• A.function is a function time (t) which has value
  0 at t 0
• The value of A at time A.time t0 is given by
  A.value A.function(t0)

29
MOST Model

• Using previous velocity to predict its current
  location as a function of time
• If the prediction is correct, the number of
  location update will be smaller
• If the prediction is incorrect??
• Another way for prediction is to use the movement
  of other objects for that area
• What is the assumption?

O
Time
30
Example MOST Model

• An object O has a location attribute L
• If it is stationary, the location attribute has
  two sub-attributes L.x and L.y
• If it is moving, its location attributes are
• L.route, L.startlocation, L.starttime,
  L.direction, L.speed and L.uncertainty
• L.route we assume that the object is moving
  along a pre-defined route
• L.startlocation a point on the route and is the
  location of the object at L.starttime
• With L.speed and L.direction, we can estimate the
  current distance of object at last update time
  t from L.startlocation on the route
• L.uncertainty its value depends on the update
  threshold and the update scheme adopted

31
Location Update with Prediction

• Under the MOST model, the results of an
  evaluation of a LDCQ is a set of tuples with each
  tuple consisting of ltobject, begin time, end
  timegt, Why?
• The begin time and end time of a tuple indicate
  the duration when the object satisfies the
  conditions of the query
• The result is determined based on its last
  recorded velocity
• The result is a prediction assuming that the
  object is moving following the information
  maintained in the location database
• The next update time of an object can be defined
  based on the begin time of the object, i.e., a
  shorter update interval is assigned if begin time
  is closer to the current time (for LDCQ)

32
Location Update with Prediction
Distance / Speed
Based on current movement, the typhoon will hit
HK at 4pm and leave at 10pm
33
Making Uses of Mobility Information

• Location updates show the movement behavior of
  the users (million users)
• What are the hot areas? Then
• Traffic conditions of the roads
• Which paths/roads people are preferred to choose
  when going from areas A to B
• Based on the location and movement behavior of
  the users, we can predict how they move in the
  future, a temporal (time) and spatial (area)
  relation
• The movement of mobile users are not totally
  random
• To reduce the number of location updates

34
Movement is affected by the connection of the
roads
35
Location Management in Cellular Networks

• Two major components in roaming (location
  management)
• Location update and Paging
• Location Update
• The mobile object informs the system its current
  location
• Every cell crossing generates an update to the
  location DB??
• Paging (broadcasting the callee ID)
• The system sends messages (to one or more base
  stations) to find a particular user
• Paging in the same cells for MSs has to be done
  one by one
• Paging delays could be very high and take a long
  delay
• Design considerations
• Large number of mobile users/objects
  (scalability)
• The importance to limit the number of location
  updates
• Limited bandwidth for paging
• Minimize and limit the delay in connection
• Where to put the location database? Distributed
  or centralized?
• Expensive uplink communication cost (minimize the
  updates)

36
Two Extreme Cases
Location update cost Vs. paging cost
Location is updated for every cell crossing (high
cost) Needs to page only one cell (low cost)
Location is never updated (no cost) Needs to page
every cells (high cost)
Location update
37
Location Area
Partition the service region into different
location areas. Each area consists of multiple
cells. How?
38
Location Update (location area)
Location update is performed when there is a
boundary crossing, i.e., into another location
area.
LA-1
How to minimize the no. of location updates How
to determine the size of an LA? How to group the
cells into LAs?
LA-2
No location update
Location update
39
Ping-Pong Effect
A lot of location updates may be generated if a
user crosses LA boundary frequently. How to
resolve this problem?
LA-1
40
The Two-Tiers Method

• It is using a location area based scheme with a a
  two-tier system of home and visited databases for
  maintaining the location information of mobile
  stations (Why? simplicity)
• When a user subscribes to the services of a
  cellular network, a record is created in the
  systems database, called Home Location Register
  (HLR)
• HLR is referred to as home system of the mobile
  objects. It is a network database that stores and
  manages all mobile subscriptions of a specific
  operator
• HLR is the location register to which an MS
  identity is assigned for record purposes, such as
  directory number, profile information, current
  location and validation period

41
The Two-Tiers Method

• Each MSC maintains a visitor location register
  (VLR)
• When mobile user visits a PCS network other than
  the HLR, a temporary record for the mobile user
  is created in Visitor Location Register (VLR)
• VLR temporarily stores subscription information
  for the visiting subscribers

42
Example Location Update
(5)
(4)
(3)
(6)
MSC
MSC
(2)
Mobile Switching Center
(1)
43
Location Update Procedures

• MS transmits a LU to the new BS
• The BS forwards the LU to the MSC
• MSC sends LU to the HLR
• HLR does the following
• authenticate the MS
• record the ID of the new VLR
• send an ACK to the new VLR
• HLR sends a registration cancellation message to
  the old VLR
• The old VLR removes the record of the MS and
  returns an ACK to HLR

44
Connection to a mobile station

• The call delivery (or call termination) procedure
  to an MS
• If a someone calls a mobile station, the call is
  forwarded to a switch, which queries the HLR to
  find the current VLR of the MS to get a routable
  address
• The VLR returns the routable address to the
  originating switch through the HLR
• Based on the routable address, a link is set up
  from the originating switch to the MS through the
  visited MSC

45
Searching a Mobile Station
46
Distance Based (N 3)
Suppose location update is performed when
entering this cell.
Location update
Comparing with fixed LA??
47
Blanket Paging
Page every cells within the LA.
48
Sequential Paging
Page the cells sequentially until the user is
found
8
9
2
7
1
10
3
4
6
5
49
How to determine paging order?

• Method 1 Shortest Distance First
• Pages the user starting from the cell where he
  last updated his location
• Move outward in a shortest-distance-first order
• Ties are broken arbitrarily
• Example (in a road)

Last Location Update
1
2
3
4
5
50
How to determine paging order?

• Method 2 Based on Location Probability
• Estimate the probability that a user is located
  in each cell within the current LA
• Page the cells in decreasing order of probability
• Example (in a highway)

Suppose Prob. Distribution is 0.05, 0.2, 0.4,
0.25, 0.1 Paging order C, D, B, E, A
Last Location Update
A
B
C
D
E
51
Paging Delay

• In our previous example, the expected delay is
• ED 0.4 1 0.25 2 0.2 3 0.1 4
  0.05 5
• 2.15 (paging cycles)
• Worst-case delay is 5 paging cycles
• The expected number of cells to be paged is also
  2.15
• Worst-case 5 cells

52
Blanket Paging vs. Sequential Paging
Sequential group paging may not be used if there
is a constraint on paging delay
53
Sequential Group Paging
Page groups of cells sequentially until the user
is found.
3
4
2
3
1
Worst case delay is 4 cycles.
2
4
2
3
3
54
Database Management

• Location Update
• Involves the updating of location databases
• Call Delivery
• Involves the querying of location databases
• The cost is very high if the MS is located far
  away from its HLR
• e.g. if the MS is roaming in Europe and its HLR
  is in HK
• Three Enhancement Methods
• Per User Location Caching
• User Profile Replication
• Forwarding Pointers

55
Per-User Location Caching

• Every time user x is called, xs location is
  cached at the VLR in the callers LA
• Any subsequent call to x originated from that LA
  can reuse this information
• No need to contact user xs HLR

56
Cache Invalidation

• Eager Caching
• Whenever a user moves to a new LA, all cache
  entries for this users location are updated
• Location update cost increases if a user moves
  frequently
• Lazy Caching
• Cache update is not performed
• Two cases can occur a hit or a miss
• In case of a miss,
• contact the HLR
• there is an additional cost, since the cached VLR
  must be visited first

57
User Profile Replication

• Observation
• Each user usually communicates frequently with a
  small number of sources
• How can we make use of this observation?
• User profiles are replicated at selected
  databases to reduce the cost of querying the HLR

58
User Profile Replication

• When a call is initiated from a certain LA, the
  corresponding MSC determines if a replication of
  the called MSs user profile is available locally
• If available, no HLR query is needed
• When the MS moves to another location, the
  network updates all replications

59
Pointer Forwarding

• Each time a user moves to a new LA, a forwarding
  pointer is set up to its pervious VLR to point
  to the new VLR
• Calls to the user will first query the HLR to
  determine the first VLR and then follow the chain
  to reach the current VLR.
• The length of the pointer chain is limited to a
  maximum value N
• This method can reduce the cost of updating the
  HLR

60
Location Database Organization

• The problem of 2-tier architecture
• Searching delay depends on the location from the
  HLR and how the cells are organized in location
  areas and managed by the MSCs
• Not scalable (centralized database, HLR)
• Multiple databases for moving object location
  management
• Organized in a hierarchical tree -gt How
• Based on the mobility of the users (objects)
• The mobility of many objects possess spatial
  (space) and temporal (time) relationship
• Affected by road/path connections
• Change gradually with time

61
Hierarchical Location DBs

• Maintain a hierarchy of location registers
  (databases)
• A location database at a higher level contains
  location information (location pointers) for all
  objects below it

1
2
62
Movement is affected by the connections of the
roads
1
2
3
4
16
63
Hierarchical Location DBs
Call
caller
64
Hierarchical Location DBs
Move
old location
new location
65
Hierarchical vs. Two-tiers
() No pre-assigned HLR () Support
Locality (-) Increased number of operations
(database operations and communication
messages) (-) Increased load and storage
requirements at the higher-levels
66
Location Area Planning
Partitions
P3
P4
P5
P1
P2
User x
User x
67
Location Area Planning
How to group the areas?
68
Location Area Planning
69
Location Area Planning

• If an MS moves from location area B to location
  area F through location area C
• The system has to insert the location information
  of the MS into the location databases labeled as
  1.2.2 and 1.2, and then
• Delete the corresponding records in the databases
  labeled 1.1.1 1.1.3, 1.1
• If an MS at location area A requests a connection
  call to the MS located at location area B
• The system has to query the location databases
  upward from the location database responsible for
  location area A to the least common ancestor
  (LCA) of location areas A and B, and then,
• Go along the proper links of the tree to the leaf
  node which is responsible for location area B (in
  which the called MS currently resides)

70
Location Area Planning

• We call the length of the path traveling from
  location areas A to B as the distance between the
  two location areas, i.e., dis(A,B)
• The location update cost and the searching cost
  depends on the distance between the starting cell
  and the destination cell in the location database
  tree
• If the distance is large, the searching cost will
  be heavy
• the distance between two location areas in a
  location database tree depends on how the
  location databases are organized in the hierarchy
• Although location area C is geographically next
  to location area F, the location update cost
  could be high for an MS moving from location area
  C to location area F since the system needs to
  update the location databases labeled as 1.1.3,
  1.1, 1.2 and 1.2.2

71
Location Area Planning

• The Location Area Planning (LAP) Problem
• An instance G (V, E, m) of the LAP problem
  consists a set of vertices which are numbered
  from 1 to n (i.e., V 1, 2, 3 n). An edge (i,
  j) in E indicates that an MT may move from cell i
  to cell j directly, and vice versa
• Each edge is associated with a weight m(i, j) to
  denote the traffic volume between cell i and cell
  j.
• A location area planning problem could be
  represented by a subset of V. The goal is to seek
  a location area design to partition V into
  disjoint subsets v1, v2, and vn so that the
  total value is minimized, i.e.,

72
Location Area Planning
The Number of boundary crossings between cells
73
Location Area Planning

• Suppose we have 6 cells in the service area under
  consideration
• Each number on the boundary between two
  neighboring cells indicates the average number of
  boundary crossings between the two cells per unit
  time
• Assume that each LA could have 3 cells in maximum
• Calculate the update cost for different location
  area designs, e.g., A, B, D, C, E, F, to
  group the cells into two LAs
• Identify the benefits (and cost) of the each
  grouping
• The benefit in grouping cells A, B, and D into
  the same group is 17 13 23 53

74
References

• Schiller 4.1.6
• Y.B. Lin and I. Chlamtac, Wireless and Mobile
  Network Architecture, John Wiley and Sons
  (chapters 2 and 3)
• A. P. Sistla, Ouri Wolfson, Sam Chamberlain and
  Son Dao, Modeling and Querying Moving Objects,
  1997 International Conference on Data Engineering
• I. F. Akyildiz et. al., Mobility management in
  next-generation wireless systems, Proceedings of
  the IEEE, pp. 1347-1384, vol. 87, no. 8, Aug.
  1999.
• V. W.-S. Wong and V. C. M. Leung, Location
  management for next-generation personal
  communications networks, IEEE Network, pp.
  18-24, Sep/Oct 2000.