Contribución al Desarrollo de Soluciones
Contribución al Desarrollo de Soluciones
Para la Integración de Métodos de
Para la Integración de Métodos de
Establecimiento de Sesión en Redes 4G
Establecimiento de Sesión en Redes 4G
(
(ContributionContribution toto DesigningDesigning SuitableSuitable Session
Session SetupSetup SolutionsSolutions in 4G in 4G NetworksNetworks))
Autor: D. Antonio Cuevas Casado
Autor: D. Antonio Cuevas Casado
Director: Dr. José Ignacio Moreno Novella
Director: Dr. José Ignacio Moreno Novella
TESIS DOCTORAL
TESIS DOCTORAL
Departamento de Ingeniería Telemática
Outlook
Design suitable session setup solutions in
4G networks
1. Introduction, Objectives 2. State of the Art
3. Integrating Sessions: Requirements
4. 4G Proposed Architecture and Session Setup
Support
5. Signalling 6. Validation
Introduction, Context
Migration of the circuit switched networks
towards an IP world is happening:
Vinton Cerf: “no sense to keep two separate
networks”
Reasons for such an important change
Increase revenue due to ease of building new
services over the Internet and their acceptance by the users
Reduce costs to operators: manage only one
Introduction, Context
For us, 4G is an Internet like network with
extra features, many, such as mobility, already present in cellular networks
A 4G network is not a 3G “All-IP” network
(3G network with an IP overlay)
Introduction, challenges and open
points
In 4G, terminals are IP hosts. Risks for the operators to become bit pipes
Migration can not be avoided, avoid its bad consequences for the operators
Convergence in business models?? still not clear. Thus…
One topic to still handle in 4G networks is session setup.
The problems come from the fact that a session is composed of subsessions (transport, application) handled by different business entities
Introduction, objectives
Design Session Setup solutions integrating
application and network layer aspects and services in 4G networks
Focus on orchestration between the
different entities, systems, modules
But keeping Internet paradigm: transport &
service separation
Position the network operator in the core of
Outlook
State of the Art
See what is profitable, research opportunities matching our
objectives
Integrating Sessions: Requirements
Sessions are split, advantages and requirements to integrate
them
4G Proposed Architecture and Session Setup Support
Design needed aspects, session setup techniques in different
scenarios
Signaling
Closes the solution, important for understanding simulations
Validation
Methods, rules to design 4G networks
Outlook
Introduction
State of the Art
See what is profitable, research opportunities
matching our objectives
Session Setup Techniques
IMS 4G
Integrating Sessions: Requirements
4G Proposed Architecture and Session Setup Support Signaling
IETF session setup solutions
RSVP: offer + answer. To setup transport
sessions. Few parameters, simple model
In the IntServ framework
In the “application” layer:
SIP & SDP (+SDPng) very adequate, plus
mechanisms (RFCs) to coordinate SIP and QoS setup
IMS
ISC Interface SIP/SDP IPv6 HSS UTRAN SGSN GGSN Home Subscriber Server P-CSCF Application Server or OSA GatewayIMS
SIP/SDP IMS signalling IP Flows S-CSCF UE PDF SIP proxy infrastructure (CSCFs) integrated with 3G networks elements (HSS, GGSN -via the PDF-)4G research efforts
Research efforts such as mobility, QoS and
AAA in IP(v6) based networks
Many existing solutions, challenge to
integrate them
Tests show that 4G can support Multimedia
applications (which are the main
applications concerned with session setup)
Some works about session setup, but
State of the art vs. thesis innovation &
objectives
IMS already fulfils our idea
Semi walled garden business model
But there is room to fit our research
objectives:
Enhance and migrate IMS it to 4G networks with many access technologies
Support “non proxied” applications (e-mail) From partial 4G solutions to a complete
framework
Outlook
Introduction State of the Art
Integrating Sessions: Requirements Sessions are split, advantages and
requirements to integrate them
Sessions in the different entities
Business models, Session Integration
Users’ Profiles
4G Proposed Architecture and Session Setup Support Signaling
Validation Conclusions
The 4 entities
They control different aspects of the
subsessions standing for a service
Conferees (e.g. caller & callee in
phone calls)
VASPs: Services providers (they are
also conferees)
Terminal dependent Parameters: Screen size, processing power… 3rd party resource dependent parameters (network): mean BW, peak BW, error rate… Mix: Codecs to employ,…
Session setup parameters
Network dependent parameters may be priced and related to users’ profiles. Which user profile should be followed when setting up sessions?: The Best, Caller’s one, …?
Business models, features to achieve
Single sign-on
Unified & non duplicated billing Service bundles, packaging
Need of aggregating sessions
See whether the next-presented business
Network operator
Trust
relationship
3rd
Party Service Providers 3rd
Party Service Providers
Trust
relationship
Trust
relationship Trust relationship Trust relationship
Internet Business model
Not possible to aggregate sessions and
thus to offer advanced features
Network operator + Service provider
Trust
relationship
Telephony Business model
Easy to aggregate sessions and offer
advanced features
3rd
Party Service Providers
Trust
relationship
Semi-walled garden business model
To aggregate sessions and offer advanced
features we to design appropriate
mechanisms in 4G networks (Thesis Topic)
Trust relationship
Network operator
Users’ Profiles
The NVUP. More complex than flat rates? Deutsche Telekom President: “create different
QoS classes”…
but “Interface” to the users simple (flat rates) A problem: Application profiles may differ
with NVUP. Example:
NVUP discount on AF traffic from 20:00 to 8:00. News Streaming: Discount from 21:00 to 7:00 One of the pieces needed to solve it: the
Conclusions
Benefits of Semi-walled garden model QoS-enabled data transport seen as
another service
Need to integrate services’ sessions Cost sharing
User profiles, NVUP; and the SVUP as a
way to integrate all the profiles
Need also to support Internet model and
non IMS applications, coexistence
Outlook
Introduction State of the Art
Integrating Sessions: Requirements
4G Proposed Architecture and Session
Setup Support
Design needed aspects, session setup
techniques in different scenarios
Systems
Orchestration
CN Internet AN
Systems
QoSB QoSB MMSP MMSP AA AAACN Internet AN
Systems-QoS
QoSB QoSB MMSP MMSP AA AAQoS System modules
Control network usage not only for QoS but also for SLAs issues e.g. police NVUP: B.E.<64 kbps QoSBrokers, Hierarchical
Push fine QoS control to the “edges”
QoSBroker: PDP (like IMS PDF) + network
management extra QoS handling possibilities
Proposal specialize QoSBroker in network management and not in user control
But still QoS tailored to users and applications thanks to
interactions with AAA and MMSP
Implicit signaling… and QoSB gets user related data
from AAA
DiffServ CN No access control, No session No need of flow-user association Internet IntServ like AN, Access IntServ like AN QoS Broker QoS Broker QoS Broker ER ER ER ER CR CR CR CR Measurement system Measurement system MMSP VASP
CN Internet
AN
Systems-MMSP
Multi Media Service provisioning Platform (i.e. SIP proxies)
QoSB QoSB MMSP MMSP AA AAA
MMSP modules
Like in IMS, hierarchical split: AN-MMSP,
CN-MMSP
Match with the QoS split
P-CSCF + S-CSCF in the AN-MMSP
AN-MMSP does the QoS SETUP and needs
to keep its state (no need in IMS since we have only QoS policy transfer) Extra
CN CN AN MMSP MMSP Internet AAA system (registrar service) QoS system AAA System VASP Users AN MMSP AN MMSP MMSP AN MMSP
CN Internet AN
Systems-AAA
QoSB QoSB MMSP MMSP AA AAAAAA modules & decentralization
proposal
Aggregates the sessions other entities produce NAS: access session, user-packet mapping
Decentralization: Not thesis main goal but it’s an important aspect to handle
Decentralize accounting and authorization: Delegation to AN-AA server:
Diameter credit control credit server. Plus accounting
consolidation
Authorization by transfer of selected parts of the user
profile (SAML). Concerns in roaming
Authentication keeps being centralized in the CN_AAA home server
DiffServ CN IntServ like AN AA AAA Internet MMSP QoS System MMSP VASP
Orchestration
Authorization or distribution of users
profiles?
Proposal: deliver profiles to MMSP and
QoSBroker. They are trusted and may
represent a big % of the total subsessions & services
AAA Client and AAA system to understand services
delivering parts of the user profile to not
fully trusted AAA clients (service providers)
Performance KO Privacy KO
Orchestration
Several possibilities to authorize transport
and services:
MMSPAAA, MMSPQoSB
MMSP obtains all the relevant user data from the
AAA and then can ask for user and application tailored QoS
VASPAAA, VASPQoSB, QoSBAAA
the AAA does not need to deliver user profile to the
VASPs who, still, can ask for QoS to the QoSB. QoSB will obtain the missing data from the AAA
Orchestration
AAA as sole network contact point:
VASPAAA, TerminalQoSB, QoSBAAA “Internet” scenarios
No entity informs the AAA about the
application, QoS based on NVUP
TerminalAAA (registration, done in every scenario)
TerminalQoSB, QoSBAAA
Thanks to the QoSB-AAA interaction we
Conclusion
Holistic approach
Solutions to give coherence to the session
setup split presented before
Many scenarios and possibilities evaluated
(complete the evaluation with simulation)
Coexistence of the several proposals and
heterogeneous environments
Outlook
Introduction
State of the Art
Integrating Sessions: Requirements
4G Proposed Architecture and Session Setup Support
Signaling
Closes the solution
Validation
Contributions summary
Standard protocols enhanced to fit our
needs and to co-operate between them
Session identification and its exchange Aggregation of sessions
Heterogeneous QoS and NAS sessions
supported
Registration
Setup in MMSP, VASP, Internet scenarios SAML allows transferring user profile to
Outlook
Introduction State of the Art
Integrating Sessions: Requirements
4G Proposed Architecture and Session Setup Support Signaling
Validation
Methods to design 4G networks
Simulation scenario
Test results for the “MN and MMSP strategies”
Scenario
The 4 CNs
CN CN CN CNScenario
1 of the 4 CNs
Internet AN AN AN AN AN AN AN AN QoSB MMSP AAAScenario
1 of the 32 ANs
CN QoSB MMSP AASystem load
Nodes processing speed calculated to
support 64 calls/s
AN-MMSP calculations for the MN strategy
1 major Test
Several measurements 2 minor tests
Call setup time…
the system is overloaded!!
Tiempos de inicio, fin y duración del establecimiento de las llamadas.
64 llamadas/s. Estrategia "MMSP" L la m a d a s ( o rd e n a d a s p o r ti e m p o d e i n ic io )
However, no node is overloaded!!
e.g. procesing time in the AN-AA…
Tiempo de procesamiento de los mensajes en un AN-AA
64 llamadas/s. Estrategia MMSP. Cada punto es un mensaje
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 12 16 20 24 28 32 36 40 44 48
Tiempo de de entrada del mensaje (s)
T ie m p o d e p ro c e s a m ie n to d e l m e n s a je ( s )
… but the AN-MMSP in the MMSP
strategy
Tiempo de procesamiento de los mensajes en un AN-MMSP.
64 llamadas/s. Estrategia MMSP. Cada punto es un mensaje
0.5 1 1.5 2 2.5 3 T ie m p o d e p ro c e s o d e l m e n s a je ( s )
System Load
No node overload but the system is
overloaded (session setup time increases)
Other tests
Minimum load
!= in setup time between the 2 strategies <=> the
AN-MMSP speed
maximum stable load
this difference is bigger (45%)
Effects of nodes on each other in a
complex architecture
Need for simulations, analytical results not
Peak of calls
Call rate low rate,
high rate, (during 1 s) low rate
Measurements
Session setup time
Call Setup time. MMSP strategy
Tiempo de establecimiento de las llamadas.
Pico de llamadas. Estrategia MMSP
0 1 2 3 4 5 6 7 8 9 12 22 32 42 52 62 72 82
Tiempo de inicio de la llamada (s)
T ie m p o d e e s ta b le c im ie n to ( s )
Call Setup time. MN strategy
…lower peak and shorter time to recover
Tiempo de establecimiento de las llamadas.
Pico de llamadas. Estrategia MMSP
1 2 3 4 5 6 7 8 9 T ie m p o d e p ro c e s a m ie n to ( s )
The CN nodes present a peak...
Tiempo de procesamiento de los mensajes en un CN-AAA. Pico de llamadas. Cada punto es un mensaje. Estrategia MN 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 12 22 32 42 52 62 72 82
…and not the AN ones
Tiempo de procesamiento de los mensajes en un AN-MMSP. Pico de llamadas. Cada punto es un mensaje. Estrategia MN
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 T ie m p o d e p ro c e s o d e l m e n s a je ( s )
But for the AN-MMSP in the MMSP
strategy
Tiempo de procesamiento de los mensajes en un AN-MMSP. Pico de llamadas. Cada punto es un mensaje. Estrategia MMSP
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 0 10 20 30 40 50 60 70 80 90 100
Tiempo de entrada del mensaje (s)
T ie m p o d e p ro c e s o d e l m e n s a je ( s )
Peak of calls
AN-CN split advantage, peak present only
in central nodes
Influence of the AN-MMSP in “recovering”:
in the “MMSP strategy” it presents a
processing time peak and this “strategy” takes more time to “recover”
Conclusion, methods to design 4G
1st holistic simulation
Many nodes interacting, analytical
approach not enough even in our small model
Benefits of the hierarchical approach (peak
of calls)
Bottle necks: nodes with many interactions
(AN-MMSP)=> reduce their load=> SIP routing at CN-MMSP
Outlook
State of the Art
See what is profitable, research opportunities matching our
objectives
Integrating Sessions: Requirements
Sessions are split, advantages and requirements to integrate them 4G Proposed Architecture and Session Setup Support
Design needed aspects, session setup techniques in different
scenarios
Signaling
Closes the solution, important for understanding simulations Validation
Conclusions
IMS evolution and enhancement towards
4G
First holistic 4G framework covering
several scenarios, entities, systems…
Semiwalled model achieved but keeping
Internet paradigm: transport and service separation
Other scenarios supported
Design decisions & approaches compared
Conclusion, avenues for research
Multiparty calls, mobility=> extend our
framework
Anonimization, pseudonyms: single sign
on vs. not tracking user activity
Conclusion, forecast for our proposal
Consumers’ preferences determine all
the aspects in the business (stop vs. non stop flights costs vs. price)
Users need to find any advantage in the
services
Our framework is a service provisioning
platform that may help create better services accepted by the users …
Or just compete in price but keeping
Contribución al Desarrollo de Soluciones
Contribución al Desarrollo de Soluciones
Para la Integración de Métodos de
Para la Integración de Métodos de
Establecimiento de Sesión en Redes 4G
Establecimiento de Sesión en Redes 4G
(
(ContributionContribution toto DesigningDesigning SuitableSuitable Session
Session SetupSetup SolutionsSolutions in 4G in 4G NetworksNetworks))
Autor: D. Antonio Cuevas Casado
Autor: D. Antonio Cuevas Casado
Director: Dr. José Ignacio Moreno Novella
Director: Dr. José Ignacio Moreno Novella
TESIS DOCTORAL
TESIS DOCTORAL
Departamento de Ingeniería Telemática