intel heterogeneous network solution brief

5
SOLUTION BRIEF Intel® Core™ Processor Telecommunications Industry Network operators have a wide spectrum of options for adding intelligence Exploding smart phone usage is putting unprecedented pressure on network operators to keep up with this insatiable demand while maintaining quality. The mobile data traffic forecasts, such as Cisco*'s projected 26-fold increase between 2010 and 2015, 1 are staggering. Besides consuming massive amounts of network bandwidth, phone subscribers expect the mass availability of new applications and services will continue unabated. Offering a solution to both data traffic demand and individualized consumption can be addressed by investigating adding more intelligence to network elements to improve network bandwidth utilization and expand service capacity. What is network intelligence? It is computing capacity used to manage and process data packets – beyond the typical routing and forwarding. This could be transrating, encoding, compression, provisioning, caching, prioritizing transactions and in a service network issuing coupons to phone subscribers who enter a store. This solution brief outlines some considerations for network operators looking for architectural strategies that better respond to the growing demand for data traffic and new services. Since the network is heterogeneous, with a mix of high and low bandwidth delivery mechanisms, solutions need to scale across a wide range of deployment scenarios. "We have a scalable processor architecture that allows network operators to add intelligence to a large Cloud-RAN, a small edge cloud or anything in between." Stephen Price Director of Marketing Intel's Communications Infrastructure Division Intel Heterogeneous Network Solution Brief

Upload: others

Post on 24-Jan-2022

4 views

Category:

Documents


0 download

TRANSCRIPT

SOLUTION BRIEFIntel® Core™ ProcessorTelecommunications Industry

Network operators have a wide spectrum of options for adding intelligence

Exploding smart phone usage is putting unprecedented pressure on network operators to keep up with this insatiable demand while maintaining quality. The mobile data traffic forecasts, such as Cisco*'s projected 26-fold increase between 2010 and 2015,1 are staggering. Besides consuming massive amounts of network bandwidth, phone subscribers expect the mass availability of new applications and services will continue unabated. Offering a solution to both data traffic demand and individualized consumption can be addressed by investigating adding more intelligence to network elements to improve network bandwidth utilization and expand service capacity.

What is network intelligence? It is computing capacity used to manage and process data packets – beyond the typical routing and forwarding. This could be transrating, encoding, compression, provisioning, caching, prioritizing transactions and in a service network issuing coupons to phone subscribers who enter a store.

This solution brief outlines some considerations for network operators looking for architectural strategies that better respond to the growing demand for data traffic and new services. Since the network is heterogeneous, with a mix of high and low bandwidth delivery mechanisms, solutions need to scale across a wide range of deployment scenarios.

"We have a scalable processor architecture that allows network

operators to add intelligence to a large Cloud-RAN, a small edge

cloud or anything in between."

Stephen PriceDirector of Marketing

Intel's Communications Infrastructure Division

Intel Heterogeneous Network Solution Brief

2

Where to Put the Intelligence?There's general consensus that the network needs more intelligence for various reasons, including:

• managing backhaul traffic

• implementing new local services

• improving the cloud computing experience

Conceptually, intelligence can be distributed or centralized, as depicted in Figure 1, but in reality, large network operators typically employ both topologies since their networks are far from homogeneous. In fact, it makes sense to place intelligence in multiple locations, as described in the following:

Macro topology: Adding intelligence to base stations allows more traffic to be processed by the edge of the network, thus potentially reducing backhaul traffic, speeding up data delivery, serving more users and enabling a new breed of local services.

Highly-distributed topology: Utilizing intelligent small cells to augment macro base stations provides a cost-effective means to increase coverage density, while optimizing the reuse of spectrum via cell splitting (spatial diversity). For instance, several small cells could be deployed in a large metro shopping mall to address the local capacity and service demand.

Highly-centralized topology: Consolidating intelligence in the core lowers network support costs since most computing resources are centrally located, and it allows for simpler and rather inexpensive edge devices, thereby reducing capital expenditures (CapEx).

A key factor in deciding where to locate intelligence is how much backhaul or transport capacity is available to the edge devices, like base stations and small cells. If connections are already bandwidth constrained (e.g., small pipes), moving to a centralized topology could further degrade quality of service because a larger percentage of traffic must be sent to/from the core network for processing. The amount of backhaul bandwidth is often dependent on the amount of dedicated fibre owned by the wireless operator.

Distributed Topology Using Smart Base StationsTo handle the future demand of more users and their growing appetite for network bandwidth intensive applications (e.g., video), many network operators are increasing capacity by adding smart base stations to the network. These devices leverage their close proximity to subscribers by caching media locally, which enables faster data delivery than handling the request at the central office or corporate datacenter. In this example, smart base stations at the edge of the network help improve the subscriber experience and reduce traffic to the network core – the industry has coined the concept of an “edge cloud” to represent the numerous deployment and use cases that a smart base station can address via the highly distributed computational resource available at the edge of the network.

Figure 1: Approaches for Adding Intelligence to the Macro Network

Highly-Distributed Macro High-Centralized

Load Balancer & Switch

Core Network

Evolved Packet Core (4G)

UMTS/HSPDA (3G)

Remote Radio Heads (RHH)

Core Network

Evolved Packet Core (4G)

UMTS/HSPDA (3G)

Small Cells

Base Stations

Base Stations

Edge Cloud

Cloud-RAN

Add

Intelligence

Add

Intelligence

Add

Intelligence

Add

Intelligence

All Optical

Fibre

Figure 2. Ubiquisys* Concept Intelligent Small Base Station

3

When small base stations have a significant amount of computing headroom, they can perform other functions that can further reduce backhaul traffic or help protect the network. For example, the core network can send compressed files (i.e., less bandwidth) to the base station as long as it has enough spare performance to decompress them before they are delivered to subscribers. High performance small base stations can perform deep packet inspection on incoming traffic in order to detect and quarantine infected content before it impacts other network elements or subscriber devices.

Despite requiring more application processing power than typical base stations, equipment manufacturers are developing smart small base stations, as pictured in Figure 2, that are relatively low cost and extremely compact. This Ubiquisys* intelligent small base station is based on the Intel® Core™ processor, which performs application, control plane and packet processing.

Public Hotspots

”Edge cloud” base station technology can be used to provide hotspot capacity in a local area, such as a shopping center, corporate building or neighborhood. This capability opens the door to interesting new opportunities, like geofencing, where the smart base station sends automatic alerts or notifications when mobile phone users enter, leave or move around a particular geographic region.

• Coupon distribution example

Using geofencing, a retailer can send a coupon to the phone of a customer walking up to the store, as depicted in Figure 3. In this case, the smart base station runs the application that distributes the coupon, which eliminates the need to send coupon requests back to the datacenter at store headquarters.

• Data caching example

While shopping in stores, consumers often use their smart phones to look up product information or look for the best

prices. For shoppers accessing the same content, like the store's web page, the smart base station can cache this data rather than having it pulled multiple times from the Internet. Since the localized data is physically closer to the handset, end users experience faster performance and network traffic is reduced, thus lowering network operators' backhaul costs.

Intel® architecture processors are exceptionally well-suited for these applications due to their large on-chip caches, multi-core architecture, fast system memory interfaces and proven interfaces to non volatile storage. These features enable a base station to quickly transmit large numbers of web pages, videos or other data content. In fact, these same traits explain why web hosting companies extensively use Intel processors in their data centers.

"Intel recognizes that intelligent small cells are a key network planning asset for service providers," said Rose Schooler, General Manager of Intel's Communications Infrastructure Division. "By extending powerful cloud-computing platforms closer to mobile devices, we are enabling a richer, more personal user experience."

Central

Office

Core

Routers

Serving

Gateway

Shopping Mall

Concourse

Coffee ShopStore

Web page

PDA

Coupon

Cell Phone

Small Cell

Small cell reduces

backhaul traffic

Internet

Figure 3. Smart Cell Usage Models

4

In simulations2 performed at Intel, data caching was shown to significantly reduce backhaul traffic, while increasing the data transmitted on the airlink, resulting in a better end user experience. This can be seen in Figure 4, where the datalink peak data rate increased from 2.1 Mbps to 7.2 Mbps, which was a 3.4 times performance increase. This performance gain was possible because the data was stored locally and was no longer limited by backhaul capacity.

Centralized Approach with C-RANA new network architecture, called Cloud Radio Access Network, or C-RAN, moves the communications signal processing back into a centrally located, virtualized base station, referred to as the baseband (BBU) unit pool. As a result, cell-sites are only responsible for radio transmission, thus they primarily consist of remote radio heads (RRHs) and antennas. The virtualized servers perform baseband processing for high numbers of cell-sites, for example thousands in a large datacenter. These cell sites are pooled resources and have the intelligence needed to support additional services, like Content Distribution Network (CDN), Distributed Service Network (DSN) and deep packet inspection (DPI). C-RANs can provide significant cost advantages in the areas of network energy consumption, construction (CAPEX) and operations (OPEX).

"We have been working on a Cloud-RAN proof of concept with China Mobile, the world’s largest operator by the number of subscribers and network size. Together, we developed a C-RAN prototype running on Intel® Xeon® processor-based servers processing TD-LTE signals," said Rose Schooler of Intel.

Distributed, Centralized or Both?Weighing the tradeoffs between distributed or centralized intelligence is interesting and worthwhile, but the reality is most large networks need both approaches. This poses a challenge because services have to be supported over the entire the network, regardless of whether they run in the core network or on intelligent base stations. In other words, services supported by relatively small processors in base stations should be no different than services running on powerful server processors.

Intel® architecture processors bridge this requirement in two ways. First, it's possible to use the same code base across different network elements – from base stations to cloud RAN servers. Software developed today can scale across a family of processors, from Intel® Atom™ processors to Intel® Xeon® processors, and very beneficial to the Telco industry, seven year long life support is available for embedded Intel® products.

Furthermore, equipment developers can transition from using discrete architectures per major workload (application, control plane, data plane) to a single architecture that consolidates the workloads into a more scalable and simplified solution. In contrast, today’s base stations use multiple processor architectures for specific purposes, like signal processing, data forwarding and application processing, and this tends to limit flexibility and increase cost. Using Intel® processors with multi-core technology, both radio signal processing and services processing can run on the same platform, which makes the revolutionary C-RAN technology even more cost-effective. Other Intel® technologies, such as virtualization, which supports dynamic resource allocation, and power management, making signal and application processing more efficient, further address costs.

Figure 4: Traffic Measurements – Laboratory Conditions

2.1 Mbits/sec

Blue = Airlink Red = Backhaul

0 bits/sec

Before Caching Is Activated After Caching Is Activated

8.4 Mbits/sec

0 bits/sec

For more information about Intel® solutions for telecommunication, visit www.intel.com/design/servers/solutions/telecom

1 http://www.cisco.com/en/US/solutions/collateral/ns341/ns525/ns537/ns705/ns827/white_paper_c11-520862.pdf2 Performance tests and ratings are measured using specific computer systems and/or components and reflect the approximate performance of Intel® products as measured by those tests. Any

difference in system hardware or software design or configuration may affect actual performance. Buyers should consult other sources of information to evaluate the performance of systems or components they are considering purchasing. For more information on performance tests and on the performance of Intel products, visit http://www.intel.com/performance/resources/limits.htm

Copyright © 2011 Intel Corporation. All rights reserved. Intel, the Intel logo, Intel Core, Xeon and Intel Atom are trademarks of Intel Corporation in the United States and/or other countries. *Other names and brands may be claimed as the property of others. Printed in USA 1011/BM/TM/PDF Please Recycle 326297-001US

Intel technologies are supported by a broad ecosystem, including the Intel® Embedded and Communications Alliance (Intel® ECA), a community of communications and embedded developers and solutions providers committed to the development of modular, standards-based solutions. More than 100 Intel ECA members provide software, hardware and tools solutions to help developers reduce their time-to-market and overall development cost. For more information, please visit www.intel.com/design/network/ica.

Adding Intelligence to the NetworkUltimately, there may be no clear answer for which is better: distributed or centralized intelligence. A likely outcome is for network operators to deploy a mix of both. For equipment manufacturers who want to be prepared in either case, they can choose one processor architecture, Intel architecture, for all the use cases.

Engineers planning future products, developing applications or migrating existing software will benefit from a wide range of Intel processors, technologies, application notes and tools. Intel processors are supported by an extensive ecosystem of development tools vendors, as well as Intel® tools, that enable programmers to more easily write fast, efficient, reusable code.