Transcript Achieving All the Time, Everywhere Access in Next

Achieving All the Time,
Everywhere Access in NextGeneration Mobile Networks
by Marcello Cinque,
Domenico Cotroneo and Stefano Russo
Presented by
Ashok Sahu
Goal
To leverage the availability of wireless
connection thus providing the “All the time,
Everywhere access” view of pervasive
computing.
How?
Leveraging communication availability - by
providing general and extensible vertical
handoff schemes.
Providing applications with a connection
awareness support.
Paper’s Contribution
 A Novel Mobility Management
architecture for NGMN (Next Generation
Mobile Networks) that consists of
1. Last Second Soft Handoff (LSSH)
2. Nomadic Computing SOCKetS
(NCSOCKS)- Connection aware
transport layer API
What is a Handoff?
The process of transferring an ongoing call
or data session from one channel
connected to the core network to another.
Types of handoffs
Reactive and Proactive handoff schemes.
Hard handoff and soft handoff.
Vertical and horizontal handoff.
Author’s approach :- Proactive handoff
based on Receiver Signal Strength
Indicator (RSSI) , a combination of Hard
and soft handoff Vertical handoff
Proposed handoff scheme
It consists of three phases
Initiation: the network status is monitored
to decide when to start a migration
Decision: Once the need to handoff is
triggered a new AP has to be selected.
Execution: the connection to the selected
AP is established.
Primary focus
 Primary focus is on minimizing unavailability
periods assuming that mobile device is in a zone
covered by Access points (APs).
 Unavailability can be caused by two kinds of
events i) handoff occurrence ii) cell overload
during handoff
 Event i) does not occur if a soft handoff scheme
is adopted. For this reason, soft handoff scheme
has been chosen.
Formula
 The define availability as
where Pr(H) is the probability that
a handoff occurs and Pr(O) is the
probability that the AP is
overloaded.
 Pr(O) has been farther defined as
where Nap is the number of
Access points, C is the maximum
average number of connections
that can be handled by APs. N is
the total number of admitted
connections.
 N can be further defined as where
Nc is the number of simultaneous
connections required and U is the
number of Mobile devices MDs
Initiation Phase
 They noted that its crucial to discriminate
between transient signal degradations from
permanent ones.
 Due to this they do not follow an initiation based
on fixed RSSI value. They argue that initiation
based on fixed RSSI value leads to a poor
availability due to transient RSSI degradations.
They experimentally confirm this.
 The probability of handoff occurrence in case of
simple fixed threshold mechanism is
Pr(H) = Pr(RSSI< Srssi).
 They follow an α-count mechanism instead.
α-count Mechanism for handoff initiation
 α-count function is a count and
threshold mechanism. It takes
the L-th measured RSSI as the
input than it is incremented by
1 if current RSSI is less then
threshold. It is decremented by
dec if current RSSI is greater
that threshold.
 Handoff is triggered if the
degradation becomes
permanent i.e. α-count
reaches a threshold α-T
Experimental proof for α-count behaviour
Decision phase
 The decision algorithm
considers only neighboring
APs.
 To APs are neighbors if their
distance d is less than a
certain value dmax
 The decision if taken based on
score criteria.
 Applications can easily specify
their requirements via
NCSOCKS API.
 AP topology is provided by a
specialized component, the
Map Server.
Proposed Architecture
 The major components
are
 Connection and Location
Manager (CLM)
 Nomadic Computing
Sockets
 Map Server
 The first two components
run on the device-side
platform whereas the
third is deployed on the
core network.
Connection and Location Manager (CLM)
 CLM is in charge of handling connections with the APs.
 It handles vertical and horizontal handoffs using
proposed LSSH scheme
 α-count parameters and score function weights are set
by applications via NCSOCKS API.
 CLM also manages information about the current
location of the mobile device.
 It is designed according to interface based approach, so
as to handle all the wireless channels through the same
interface, which provides several common operations as
searching APs, creating/destroying connections, building
an IP interface upon the wireless media and getting
current RSSI, delay, bandwidth and cost values.
Map Server Responsibilities
 Responsible for accepting map requests from
MD’s CLM.
 Map server has to recognize topology changes
and provide last updated information to MD’s
 To avoid bottle neck APs have been grouped
and there is a Map server for each cluster.
 Clusters should be set up taking into account the
physical topology.
Achieving connection awareness: the
NCSOCKS
 NCSOCKS use both IP communication facilities
and information gathered from data link layer,
through the CLM.
 Information flows from/to the application to/from
the CLM, via the NCSOCKS interface.
 Applications can set there QOS requirements
and can request current connection status.
 The current symbolic location, the wireless
technology and the cost are provided by the
Map Server, knowing which AP is currently being
used.
CLM and NCSOCKS class diagram
Testbed description
Pr(H) with different schemes
Pr(O) estimation with different handoff
schemes
Conclusion
 Experimental results have demonstrated that the
proposed handoff scheme reduces unavailability
periods, due to transient signal degradations
and AP overloads, significantly.
 I think it’s a useful contribution but there might
be better ways.
 Might suffer from Map server update problem.
 Session layer approach might be better.