Escolar Documentos
Profissional Documentos
Cultura Documentos
Of ta
dintgork Dis oa w
Ne
ry cove
and
www.greenpacket.com
APPLICATION NOTE
Abstract
The arrival of varied smartphones and devices has inevitably encouraged greater data consumption. Users are now more sophisticated and demand nothing less than best network performance and willing to switch providers at the slightest dissatisfaction. Smartphone generally accounts for higher ARPU and represents potential new revenue streams for enabling quad play services. Operators are aggressively addressing these challenges and improving the quality of their solution to retain existing customers and offering better service experience to build loyalty. WiFi adoption as data ofoading is gaining momentum as it improves user experience while lowering the cost of service providers. Correctly implemented, it can ease up to 20% of trafc with substantial impact of freeing up the spectrum and load balancing. Ofoading data to WiFi hotspots is economically attractive, because many carriers already operate a substantial number of hotspots. In this paper, we will examine how dynamic data ofoading can be further optimized with 3GPP based Access Network Discovery Selection Function (ANDSF). ANDSF describes how the inter-system mobility between 3GPP systems and non-3GPP systems (ie: WLAN, WiMAX, CDMA) policies and priorities can control the conditions for which a device (UE) connects to which wireless network. Operators have the exibility to manage context information in network discovery and selection procedures in push and/or pull mechanism with a policy management and control component as network-based and client-based solutions. The client-based implementation is described with reference to applying real-time connection state and policy control. Policy enforcement and policy management is mapped more accurately with an overall knowledge of the network at any given point of time. By intelligently managing the distribution of network load, congestion can be minimized.
APPLICATION NOTE
APPLICATION NOTE
Contents
Overview Dynamic Data Ofoading a Study Model - WiFi or Femtocell - Why Introduce Dynamic Data Ofoad (DDO) - Data Ofoading Tool - WiFi Data Ofoading Access Network Discovery and Selection Function - Where is the Intelligence in Dynamic Data Ofoading ANDSF Implementation - Policy Control with ANDSF Values of Dynamic Data Ofoading Capability with ANDSF - Prioritization of Access Based on Network Availability - Time Based Policy Control - Application Policy Control - Power Management - Trafc Routing Conclusion Ofoad Early, Connect Intelligently References 14 15 16 12 10 07 01 02
APPLICATION NOTE
Overview
Mobile networks are inherently constrained with the availability of spectrum and bandwidth to support the rapidly growing demands of mobile data. In a recent report by Cisco, it is estimated that global mobile data trafc growth will increase by a factor of 26 times between 2010 - 2015. The way we interact with the internet is changing. The arrival of data-centric devices (eg: iPhone, Android , Blackberry), lower subscription prices, easy access to applications is driving data consumption and content like never before. Even with operators implementing transcoding methods to adapt smaller screen sizes, for pictures and videos , operators still struggle to satisfy data demands, managing congestion and improving subscriber experience. WiFi ofoad is an alternative method to ease the burden of bandwidth capacity. However, the methodology of ofoading can be made either tightly-coupled or loosely coupled to the core network. In the tightly-coupled method, the network operator retains visibility and control over the trafc ofoad related to the location where trafc streams are ofoaded or bypassed over the network operator and relevant policies provisioned. Overall data ofoading implementation can be more secure and comprehensive by complementing i-WLAN framework with dynamic data ofoading. Further information to i-WLAN can be referenced in a previous whitepaper titled Managing Data Ofoading Securely over WLAN access networks via I-WLAN. Within the loosely-coupled method, data ofoad is unmanaged. By forcing the user onto WiFi, the network operator has no visibility on the user trafc and not subject to the operators policies. The user experience associated to the accessibility of service for example to web browsing, online shopping, downloading music or streaming video will be at risk of un-trusted network attacks such as disclosing or tampering of user details. Other issues pertinent, which may involve some form of credit card purchase. Such transactions are not subject to operator controlled service rules since the user subscription relationship is decoupled from its control policy. The applied access rules previously enforced within the walled-garden parameters will be lost once the IP trafc bypasses the packet data network (PDN) en-route to the Internet directly. These IP trafc is identied and routed purely on the IP address regardless of the policies enforced on the service application level.
APPLICATION NOTE
Source: The Cisco Visual Networking Index (VNI) Global Mobile Data Trafc Forecast Update 2011
01
WiFi or Femtocell
Given that spectrum is a premium limited resource, operators are turning to WiFi or femto-based ofoad to ease trafc needs. The stark difference between WiFi and femtocells lie in the spectrum utilization. WiFi operates in unlicensed band with a larger bandwidth capable of supporting data rates of 11Mbps up to 54Mbps (peak theoretical) and couples the backhaul onto the xed line, effectively freeing-up the 3G spectrum and improving capacity. Femtocells on the hand, operates in licensed band wirelessly over the shared air interface of 3G network. Femtocells can improve indoor coverage, but not increase capacity. WiFi ofoad improves capacity at a fraction of the cost, since there is already WiFi infrastructure readily available. Dynamic Data Ofoading can be initiated with an intelligent connection client in the device (UE). The device (UE) is able to determine the subscriber state early in the network and exchange information with the policy controller in real-time to adjust network connection state.
APPLICATION NOTE
02
APPLICATION NOTE
03
APPLICATION NOTE
04
Site Count
40 35 30 25 20 15 10 5 0
3G
Supports up to 20 active users per site
WiFi
Supports up to 45 active users per site
Figure 2: WiFi and 3G Site Count Study for average user throughput of 512kbps.
In terms of the coverage, a 3G cell range is up to 5 times of a typical WiFi cell range. Within a dense urban morphology, a 3G cell range can vary between 1-2km. (cell ranges are inter-dependent on radio planning parameters such as various modulation schemes, terrain, density of subscribers, achievable speeds. This assumes the average vehicular speed of 30km/hour). On the contrary, WiFi cell range vary between 100-200m, equating to more WiFi APs per km2 coverage radius. Despite WiFi cell range of 5 times less than a 3G node B (for every 1km, 5 WiFi APs are required), the trade-off of cell capacity is far greater. If for example the density per area is 1,000 users (assumes single device (UE), single user)per 1km2 radius with 70% active users; the 700 active subscribers requires up to 35 3G cell sites to support. However, with WiFi the number of cell sites is 55% less at 16. A typical WiFi AP cost is around USD100-200, while a 3G node B is USD30,000. For an equivalent number of cell sites, the CAPEX savings are signicant. Instead of deploying more node Bs, operators can leverage on the WiFi network within the vicinity as means of extending its coverage and relieving heavy internet trafc during or off peak hours.
APPLICATION NOTE
05
Increasing Cost
Increasing Capacity
Figure 3: Cost Comparison of WiFi AP and 3G Node B.
WiFi architecture exhibits signicantly higher data rates than 3G and it could be easily modied to increase the number of connection to open or partner WiFi network with the presence of a WiFi AP. 3G provides continuous connectivity over further cell range, with lower data rates and relatively high cost while WiFi is intermittent with high bursts of data (other consideration excluded in this study includes the impact of backhaul, leasing of transport network and installation works). With better control over data ofoad solutions, such as dynamic data ofoad via Access Network Discovery and Selection Function (ANDSF), operators can understand the user state and commensurate network state contributing to congestion. Most importantly, it detects the source of congestion trigger and resolves the issue early, before it affects the service experience. The intelligent ANDSF enabled connection client can push connection status with granularity such as location corresponding to cell ID or SSID to the Policy Charging and Rules Function (PCRF). The PCRF perform real-time policy response network state at applications to identify the congestion point. Each service stream can be easily identied by the IP address where it originates and the destination IP. By having a direct connection to PCRF, the policy information is pulled by the ANDSF connection client to determine whether a handoff is necessary to maintain the connection as applied in the examples above.
APPLICATION NOTE
The PCRF server in this instance is 3GPP ANDSF standard compliant. ANDSF module can exist independently or implemented within proprietary policy control and management server that may comprise of several functional modules, depending on the network architecture.
06
APPLICATION NOTE
Policy management and control component is used interchangeably with PCRF in this application note.
07
Connection Manager
Policy Controller
Internet
Mobile Broadband (3G UMTS) Each Network Policy includes: Access Technology Type (3G, WiFi) Access Network Priority Location
APPLICATION NOTE
08
Default connection at RAN may not always be most optimal at a given point in time from core network connectivity, the end-user, application or service perspective. However, intelligent connection managers allow it to search and prioritize the best available connection based on pre-dened requirements such as signal strength and operator policies. In this manner, intelligent resource management provides a cost-effective way to support the data trafc by optimizing network radio resource, balancing the network load, reducing network congestion and delaying CAPEX for future upgrades. ANDSF may provide a list of access networks available in the vicinity of the device (UE) for all the available access technology types requested by the device (UE). Instead of unnecessary background scanning, the ANDSF connection monitoring may be carried out to handover or perform cell selection/re-selection measurements in which the device (UE) monitors the received signal strength indicator signal levels (RSSI) received from different base stations. This reduces the battery drain on the device (UE) without continuously scanning in the background. By selectively triggering the connection of device (UE) to a given threshold or the preferred available access network types based on inter-system mobility policies, the network load can be balanced or spread out to the availability of network radio resource at the point where congestion is likely to occur. Once the device (UE) receives the policy relevant to location and validity of time interval, the ANDSF will apply the current conditions to match the policy. Policies can be tailored for different access networks, regardless of devices (UE) to enable seamless connectivity experience. In the longer term, effective access network monitoring can reduce the pressure of CAPEX upgrades to sustain the increased capacity. The added intelligence of ANDSF brings benet to the negotiation of connection state by various operator policies supported as Location-Aware Policy cell ID or associated SSID is checked against the policy database for specic locations that are preferred/relevant Access Network Discovery Policy list of available access networks(RAT type) that are available within the vicinity and facilitation of prioritized access network selection for network handover Time-based Policy conguration of policy validity time to live interval to current location and prole Subscriber-based Policy dynamic allocation of subscriber package and prole
09
ANDSF Implementation
Policy Control with ANDSF
Although operators are increasingly looking at using WiFi for ofoad in congested areas, it risks losing visibility over trafc policies that were congured for the user once it routes through WiFi. What is lacking is a way for the network to communicate to users (applications and/or websites they are using) a real-time or predicted measure of the networks congestion levels. ANDSF keeps a direct relationship to the core network by providing dynamic provision of information to the device (UE) for network discovery and selection procedures related to non-3GPP accesses; the ANDSF connection client can pull and/or with a combination of pull-push may be supported with the policy charging and rules function (PCRF). ANDSF complements both the PCRF and Deep Packet Inspection (DPI) module. By means of direct device (UE) communication at applications level towards the PCRF, the ANDSF provides indication to the device (UE) on the selection of network by reducing over the air processing required in the device (UE) to establish the wireless link.
Application Server
ANDSF client in CM
GGSN / PDSN
Internet
3GPP Access
3G Network WAG
Non-3GPP Access
APPLICATION NOTE
10
Detecting users, locations, applications or device (UE) type is crucial in determining the conditions for triggering ofoad, and under what circumstances. Increasingly, applications and devices (UE) are becoming more aware of their context what network they are connected to, what is the state of performance of the network and so forth for a given time and location. The ANDSF connection client pull subscriber state information (location based cell ID, precongured policies and time congurable connection policy), from the PCRF. The ANDSF connection client will check the existing policy. If the policy is alive, it applies the current policy. Otherwise, the ANDSF connection client will request for a new policy. Similarly, the business rule engine in the PCRF pushes dynamic and subscriber specic policy decisions by checking the subscriber specic data (eg: subscriber package, access entitlement/restriction) accordingly. Based on this information, PCRF dynamically modies the connection policy rule depending on the type of policy control required, acting as a policy management point. The policy decisions are delimited by validity of the location, time of day and discovery information. The ANDSF connection client is an intelligent module that learns of subscriber state and network state parameters over time to detect the corresponding bearer (WiFi, 3G, WiMAX) to the suitability of context (service, device type) connection. These parameters ensure that the operator retains visibility of which IP trafc are critical and retained on the 3G or WiFi connection. Through connection shifts based on context information and by combining knowledge about real-time network levels, applications and devices (past data usage history), gives greater control on ofoad strategy rather than a single isolated measurement of the radio conditions. By re-prioritizing subscribers on a congested cell site or ofoad all mobile data trafc or laptop to WiFi, during peak congestion periods, specic times of day, days, weeks or combination of them, operators can have the exibility of controlling bandwidth, QoS availability and application-specic service corresponding to service level agreements.
APPLICATION NOTE
11
APPLICATION NOTE
12
Power Management
The network discovery and selection has been an issue from a device (UE) perspective, because of battery consumption and turning on and off two radios simultaneously. With real-time network congestion state information from ANDSF server, the ANDSF connection client may employ the policy to avoid unnecessary background scan by the device (UE) to prolong battery life. Different access networks have different effects on the speed and connection establishment time that affects the intensity of battery consumption. The access network selection process can be made more efcient with the ANDSF ordered access network list.
Trafc Routing
Operators routing trafc directly to the Internet as a result of ofoading, loose the visibility of control over the subscriber policies. With a managed data ofoad approach, operators can have visibility of IP trafc which is routed over WLAN (eg: best effort internet trafc) and which IP trafc is maintained on the 3G link (eg: VoIP). The combination of PCRF, DPI and ANDSF to conduct granular data ofoad, ensures trafc routing across the network allocated to particular services can be optimized for specic bearer types. By having the visibility of IP trafc, operator can analyze user behavior related to the bulk of IP trafc generated. The added intelligence can be applied in user segmentation and creation of new service streams to segregate high ARPU potentials, moderate users and low bandwidth users.
APPLICATION NOTE
Application policy control is part of the ANDSF standard. The deployment is proprietary, depending on the implementation of the ANDSF server and network architecture. Greenpacket ANDSF client is currently testing the functional attribute with select vendors. (vendor names are withheld due to non disclosure agreement in effect)
13
Conclusion
Wi-Fi is suited for data ofoad as echoed in AT&T and PCCWs business model. Today millions of subscribers are already using WiFi as their primary source for data/internet access whether as part of their subscription plan or a free service. As observed, WiFi has played a role in relieving congestion, but at the same time generating more usage. The implementation of Dynamic Data Ofoading via ANDSF can further enhance the both the 3G and WiFi experience. Dynamic data ofoading is necessary to minimize congestion. The earlier a congestion trigger is controlled and managed the operator can continue to improve the distribution of network resources from the perspective of spectrum planning. Since networks do not behave identically nor follow the same trafc model, data ofoading exercise can facilitate balancing the load whenever congestion occurs. Policy enforcement and policy management is mapped more accurately with an overall knowledge of the network at any given point of time, rather than a silo representation at the aggregation point at the backend. More and more operators are embracing data ofoading as means to satisfy the bandwidth requirements. With dynamic data ofoading methodology, the operators can capitalize on extending its network, while maintaining a balanced network investment.
APPLICATION NOTE
14
Free Consultation
If you would like a free consultation on how you can apply data ofoading with ANDSF, feel free to contact us at marketing.gp@greenpacket.com. Kindly quote the reference code, SAP0611 when you contact us.
APPLICATION NOTE
15
References
1. The Cisco Visual Networking Index (VNI) Global Mobile Data Trafc Forecast Update 2011. 2. The Role of Adaptive Policy Control and Smart Caps in Managing Mobile Data Growth. 3. 3GPP TS 23.402 V8.0 3. Bearer-aware policy management and charging, Disruptive Analysis, Dean Bubley.
APPLICATION NOTE
16
San Francisco Kuala Lumpur Singapore Shanghai Taiwan Sydney Bahrain Bangkok Hong Kong
Associate Member
Copyright 2001-2011 Green Packet Berhad. All rights reserved. No part of this publication may be reproduced, transmitted, transcribed, stored in a retrieval system, or translated into any language, in any form by any means, without the written permission of Green Packet Berhad. Green Packet Berhad reserves the right to modify or discontinue any product or piece of literature at anytime without prior notice.