Registration-Area-based Location Management
Dec 30, 2015
Location Management: ContextMobility Management: Enables users to support mobile users, allowing them to move, while simultaneously offering them incoming calls, data packets, and other services. Consists of:
1. Location management: tracking mobiles and locating them prior to establishing incoming calls (deliverying pending messages).
2. Handoff management (a.k.a. automatic link transfer): rerouting connections with minimal degradation of QoS.
Location Management in Personal Communication Systems (PCS)
Keeps track of the location of each user (location tracking)Used in call delivery and in location-dependent servicesCells organized in clusters called Registration Areas (RAs)One Location Register for each RA.One Home Location Register for each user.When current LRHLR then it is called Visitor Location Register (VLR).
MHBS
MSS
Location Management Problem
In static networks, a terminal’s network address serves two purposes:
1. End-node identifier2. Location (access point) identifier
Location management keeps mapping between an end-node identifier and its location identifier Basically a directory problem.
Registration Area-Based Location Management
Registration area == a group of cells; update only when crossing a registrationarea boundary
Location Management: Basic Operations
Two primitive operations:1. Lookup (a.k.a. search/find/paging/locating) operation: is
the procedure by which the network finds the location of the mobile. required when a call (message) to a user is placed (to be
delivered)
2. Update (a.k.a tracking/move/registration) operation: is the procedure by which the network elements update information about the location of the mobile. required when a user changes its “location” The information gathered during updating/tracking is used
during the locating operation
Lookup/Update in PCS
Lookup: when a call is placed from cell i to user x, the VLR at cell i is queried first, and only if the user is not found there, is x’s HLR contacted.
Update: when user x moves from cell i to j, in addition to updating x’s HLR, the entry of x is deleted from VLR at cell i, and a new entry for x is added to the VLR at cell j.
Location Management
infrastructure-based mobile networksmaintains location information of mobile elementsLocation information is used to properly route data and calls to mobile elementsTwo basic operations Location information update (set info) Location information search (acquire info)
Cost-based description Cost, as a metric of resource use.
Load of Location management
L=fs cs + fu cu
LocationManagement
execution costs
Updateoperations
Searchoperations
fu fs
cu cs
rates
Location Management: Schemes
Several schemes have been developed which are motivated by fundamental trade-off between search operation cost and update operation cost.
Schemes which try to minimize one cost tend to increase the other cost
Try to optimize the aggregate cost or normalized cost.
Categorization:1. Update Scheme: Static or Dynamic
Static update scheme: registration areas Dynamic update scheme: distance/time/movement based strategy
2. Locating Scheme: Static or Dynamic Static location scheme: page all the cells in the network Dynamic location scheme: expanding ring search centered at last
reported location of the the user
3. Database Architecture: Flat or Hierarchical
Selection of LM Schemes
Cost of location updates and lookupsMaximum service capacity of each location database = the maximum rate of updates and lookups that
each database can service
Space restrictions (size of the location database)Type and relative frequency of call to move operations (call-to-mobility ratio (CMR)).
Example basic scheme
Never-updateZero update costWill have to search the entire network
based on the last-known location
Always-updateNo search cost
Distance-based updateTime-based update
LM improvementsUpdate/search trade-off optimizations
Principle of update/search trade-off The more effort spent in updating
the information, the less effort needed to seek the information
And vice versa Redistribute rates of updates and
searches to overall load reduction
L = fs cs + fu cu ↓L =
↓fs cs + ↑fu cu Forwarding Pointers, Location Caching etc.
Characteristic: optimal point depends on call to mobility ratio(CMR = fs / fu )
Non-trade-off optimizations Do not conform to the update-
search trade-off principle Unilateral reduction of one (or
more) load components
L = fs cs + fu cu ↓L
= ↓fs cs + fu cu Predictive registration, predictive
paging Characteristic: optimal point
depends on knowledge of the terminal mobility and/or call model.
RA overlapping as LM improvement
Can contain more of a mobile terminal’s mobilitya
Less registrationsa,b,c
eliminate registrations of “border movements”b
Load balancinga
Minimize call lossd
aOkasaka and Onoe, Proceedings IEEE VTC’91bMarkoulidakis et al., ACM/Baltzer Wireless Networks 1(1):17-29, 1995dBejerano and Cidon, Proceedings ACM MobiCom’98cWang and Akyildiz, Proceedings ACM MobiCom’00
Intra and Inter-RA Handoffs
Let mobile m be in cell c and registered in RAk (c RAk ). move from cell c to cell d.
In any configuration, we define two types of handoffs: Intra-RA handoff: if d RAk.
Inter-RA handoff: if d RAk. after handoff the mobile is registered in RAl such that
d Core(l).
Intra- and Inter-RA Calls
An intra-RA call is one in which both the caller mobile and the callee mobile are in the same RA.
An inter-RA call is one in which both the caller mobile and the callee mobile are in different RA.
Non-overlapping & Overlapping RAs
Inter-RA hand-off: a user changes cells and RAsIntra-RA hand-off: a user changes cells within an RA.Inter-RA hand-off doesn’t happen as long as the hand-off can be intra-RA.A non-overlapping cell is serviced by one LR. A overlapping cell is serviced by multiple LRs. Reduction of inter-RA hand-
offs.
No overlapping
With overlappingA
B C
CB
A
12
3
12
3
Registration Area Overlapping
Advantages: Each RA can provide service to more mobiles within their
covered area. Reduces the number of inter-RA handoffs Reduce the load to update mobile’s HLR.
Disadvantages: the communication overhead for call-delivery and intra-RA
handoff is increased. the increase in overhead depends upon the underlying
network topology. If this overhead is ignored then the extreme configuration in
which each RA has all the cells in the system becomes the “optimal” configuration.
Proposed Scheme
Dynamically adapts the registration areas to the aggregate call and mobility pattern such that
1. the expected update overhead on mobiles decreases
2. the expected overall signaling system load does not exceed a predefined limit.
Some Notations
Core(k): set of cells directly connected to MSSk.
RAk: Registration Area of MSSk. a dynamic set of cells.
Configuration C = {RA1, RA2, ... , RAM}.
Reconfiguration: changing of configuration The MSSs periodically reconfigure the system in a
distributed manner at fixed interval of time T.
Permissible ConfigurationsProperty 1: An RA has at least one cell. no RA is empty i.e. all MSSs are used.
G = (V,E): cell adjacency graph V: set of all cells in the system E: (v1,v2) E iff cells v1 and v2 are neighbors.
Property 2: The subgraph of G induced by any RA is connected. no RA has any group of cells disconnected from the
remaining cells.
Property 3: RAk Core(k) = Core(k). Initially, RAk = Core(k).
Effect on Intra-Handoff and Call Delivery Costs
Even though mobiles a and b belong to the same RA, any calls between them would need to go through two MSSs.
Dynamically Resizing RAs
We need to find optimal configuration (allowing overlapping RAs) i.e. configuration which minimizes load on MSSs.When move and call patterns periodically change, a static scheme may not provide a good solution.Our Approach: Allow RAs to be dynamically adapted.Periodically resize RAs to minimize MSS load: Resizing criterion: load reduction due to lesser number of
inter-RA handoffs > increase in load due to more expensive call delivery and intra-RA handoffs.
If resizing criterion is ignored then each RA will grow to maximum size.
Inclusion and Exclusion Boundary
In order to facilitate orderly growth and shrinking of RAs, an MSS only includes and excludes cells from its RAs current boundary.Two types of boundary:
1. Internal Boundary2. External Boundary
MHBS
MSS
Problem Formulation
Assume an initial non-overlapping topology and configuration Assumption for greater applicability Still applicable to topologies that inherently support
overlapping
Find the inherent impact of a cell inclusionKeep track of cell-to-cell hand-off and call rates, with respect to users of one registration area:
– nc(r,i,s,j) , nu(r,i,s,j) Use those rates with the inherent impact of cell inclusion to
see if it is beneficial to include a cell
Restrictions Cannot exclude cells that initially belonged to the RA
Core of RA
Inclusion/Exclusion Decision
The decision to include or exclude a candidate cell is based on whether the resulting configuration will have a lower expected load on MSS.For a given system configuration A, mobility pattern M, and call C, SystemLoad(A,M,C) is the combined signaling load (in terms of message time complexity) as a result of all the handoffs due to M and call-deliveries due to C: SystemLoad(A,M,C) = Load(k,M,C).In case of inter_RA handoffs and call-deliveries we spilt the signaling overhead equally between the two MSSs involved.
What changes when a cell x is included to an RA r
mobility to the cell x from cells of the RA r is now intra-RA mobilitymobility from the cell x to the rest of the RA r performed by users already registered in r is now intra-RA mobility.calls to x from cells of r are now intra-RA calls.calls from users of r that are in x to rest of r are now intra-RA calls.Mobility of users in r that move out of cell x into a new RA is now inter-RA mobility.Inter-RA calls of users in r that call from cell x is inter-RA call loading to r.
Call the decreasing part of the load Costinc and the increasing part Costdec.
Now intra-RAhand-offs
Here inter-RAhand-offs
Now retainingprevious RA
New inter-RAhand-offs
RA border
HereInter-RA calls
NowIntra-RA calls
NewInter-RA calls
Mobile terminalsregistered in RA
Mobile terminalsnot registered in the RA
Finding the signaling costs
There are three cases of signaling costs in a quadruplet (r,i,s,j) 1. Cells i, j are both core
cells2. One cell is core cell, the
other is non-core cell (included)
3. Both cells i, j are non-core
Assuming a typical signaling protocol derive the costs for each operation
av
ac
vc
vv
Algorithm
Outline1. Calculate the cells along the RA
periphery– Both inside and outside the
periphery2. Send data requests to those
cells3. Receive data responses from
cells– Those values nc(r,i,s,j) , nu(r,i,s,j)
4. For each cell, calculate if it should be included or excluded
– Make sure that there are no holes created in the geographical continuity of RA
5. Send inclusion or exclusion cells
Calculate border cells
Query border cells for statistics
Calculate what to include and what
to exclude
Notify cells of their inclusion or
exclusion
LR cells
Simulation
Discrete Event Simulation using SES WorkbenchMobility Models:
1. Highway 2. Urban
1. 1-Dimensional Random Walk 400 cells divided in nine RAs 10,000 MHs call frequency: 1 call per hr to 1 call per 15 min move frequency: 1 handoff per hr to 1 call per 15 min
2. 2-Dimensional Random Walk 50x50 cells organized as 5x5 RA.
Urban ModelUrban systems (localized traffic): Overlapping does helpUses a random walk model transition probability of a MH moving
farther from a pre-defined pole of attraction drops progressively with the distance from that pole.
Effect of CMR to load with respect to overlapping (1)
100000
1e+06
1e+07
1e+08
0.1 1 10
Lo
ad
(C
os
t) o
f e
ac
h r
eg
istr
ati
on
are
a
CMR (achieved by varying call frequency)
Cost of LM for various CMRs keeping move rate constant
overlapping(0)overlapping(1)overlapping(2)overlapping(3)overlapping(4)
Effect of CMR to load with respect to overlapping (2)
100000
1e+06
1e+07
0.1 1 10
Lo
ad
(C
os
t) o
f e
ac
h r
eg
istr
ati
on
are
a
CMR (achieved by varying move frequency)
Load on a LA for various CMRs keeping call rate constant
overlapping(0)overlapping(1)overlapping(2)overlapping(3)overlapping(4)
Impact of choice on subsequent registrationsand impact of knowledge of mobility
Without probabilities A mobile terminal can move to either RA 1 or
RA 2 Assume that the registration choice is RA 2 However, the mobile terminal moves to the upper end of RA 1,
and has to perform one more registration Bad choice, there will be two registrations total
With probabilities 0.3 probability of moving to RA 1 0.7 probability of moving to RA 2 Then we always choose the most probable On average:
70% of the times the choice will cause 1 registration 30% of the times the choice will cause 2 registrations 1 x 0.7 + 2 x 0.3 = 1.3 registrations
Concept of regions: We will refer to an overlapped portion as “region” Availability of registration areas is the same throughout a
region Performing a second registration within the same region has
no benefit
RA 1
RA 2
0.7
0.3
Off-line FormulationWe need to find the minimum number of registrations along a mobility pathAssume a mobility path pfrom regions ‘a’ to ‘i’. (a)We can make a graph of region adjacency. (b)At each region we know the availability set A, the set of available RAs in the regionUsing a sort of inner product of mobility path p and availability sets, we make a new partially ordered directed graph. (c)
If an edge goes from a node (k, R i) to (m, R i), then it has weight 0
If an edge goes from a node (k, R i) to (m, R j), then it has weight 1
Problem is now defined as shortest path problem
Can be solved in linear timeM(r, p): minimum number of registrations along path p with initial registration choice r
h
g
d
cb
a
e f
i
r1
r6
r5
r4
3r
r2 r2
r1
r6
r4
3r
r5
(b)
e
h
i
(a)
g
d
cb
a
f
a b c d f g he i
R1
R2
R3
R4
R5
R6
(a ,R1) (b,R1)
weight = 1
weight = 0
(b,R6)
(c ,R1)
(c ,R2)
(c ,R6)
(d,R2)
(d,R6) (e ,R6) (f ,R6)
(e ,R5)
(f ,R2)
(f ,R3)
(f ,R5)
(g,R2)
(g,R3)
(g,R5)
(h,R2)
(h,R3) (i,R3)
(d,null)(a ,Ro=R1)
ts
G GM M
(e ,R2)
(c)
On-line FormulationOn-line computational model
Need to make a decision at every new region Decision choices include retaining the registration
choice so long as it is available.
Knowledge of mobility Assumed to be given as a random walk graph
With respect to walking across regions Can be extracted from statistical data.
Problem formulated as: Find the averaged effect of a registration choice r to
all probable subsequent paths p of length k, starting at region g:
Expct(r,g,k) = all paths p Prob[p] × M(r,p) Length k is referred to as look-ahead depth
Algorithm Straightforward computation of Expct Finds a registration that minimizes Expct Exponential complexity with respect to look-ahead
depth
Paths of
length 1
Paths of
length 2
Paths of
length 3
Paths of
length 4
Analysis of CompetitivenessCompetitiveness ratio of an on-line algorithm
Defined with respect to a measure on a solution The maximum possible ratio of the measure of the
algorithm’s solution over the measure of the off-line optimal solution for an arbitrary input
Competitiveness is usually proved using an adversary approachExample for 2-competitive:
There are three RAs: A, B, C For any move, we can move to an overlapping region
not covered by the chosen RA Example on-line:
(AB)B(AC)C(AB)A(BC)C(AB)B(AC)A Example off-line
(AB)A(AC)(AB)B(BC)(AB)A(AC)
Can be expanded to n-competitive for any arbitrary n.
Unlimited lower boundHowever, for any given topology, there is an upper bound to the competitive ratioa
(AB)
(AC)
(BC)B
CA
aKonjevod et al., ACM DIAL’M 2002
Average Case ComparisonDetermine probability that The Expct-minimizing algorithm matches the
performance of off-line optimal The random approach matches the performance of the
off-line optimalAnalysis shows that there is a certain probability that: Random choice will match the optimal algorithm, Prandom ORS algorithm will match the optimal algorithm, Pexpct
Comparison of those values shows that– Pexpct ≥ Prandom
ORS algorithm is more likely to match the optimal algorithm
Minimizing Hard registrations
Soft registration model Assumption: registrations while not on a hand-off are cheaper (soft
registrations) Objective: Minimize hard registrations.
Algorithm Pre-emptive application (use before we need) Minimize Expct(1), i.e., look-ahead 1 Converts hard registrations to soft registrations
How cheap a soft registration can be to still have a benefit We use the load function to investigate If a is the increase in total number of registrations
and b is the average update cost reduction (because hard registrations are converted to soft) Then:
Analysis shows linear relation to registrations: The cost ratio between soft and hard registrations has to be at greater than
the ratio of increase in total number of registrations.
ba
cfcf
bcafcfcfcf
L
LL
uuss
uussuuss 10
)()(0