Smoothing mobile broadband peaksAdaptive content optimization in <strong>the</strong> RANby Scott Hilton, VP and GM, Broadband Optimization Solutions, Sycamore NetworksMobile operators across <strong>the</strong> globe are struggling with network congestion caused by <strong>the</strong>unprecedented uptake of mobile broadband. <strong>As</strong> well as baseline traffic, high demand trafficpeaks have increased significantly and can have a major impact on <strong>the</strong> radio access network.Adaptive content optimization is a compelling <strong>so</strong>lution, providing an effective mechanismfor relieving traffic congestion and managing <strong>the</strong> growth of mobile broadband traffic.Scott Hilton is <strong>the</strong> Vice President and General Manager of <strong>the</strong> Broadband Optimization Solutions business unit at Sycamore Networks.Before joining Sycamore, Mr Hilton was Vice President of Product Marketing and Management at 3Com Corporation, and prior to3Com, Vice President of Product Management for <strong>the</strong> IP Services division of Lucent Technologies.Scott Hilton completed a B.S. in Electrical Engineering from Duke University and an M.S. in Electrical Engineering from George Ma<strong>so</strong>nUniversity.In <strong>the</strong> first half of 2010, <strong>the</strong> average US mobilesubscriber consumed an average of 230MBof data per month, a rise of 50 per cent over<strong>the</strong> previous six months. The increased useof smartphones and mobile broadband hasclearly driven this growth, with 31 per centof <strong>the</strong> US subscriber base now classified assmartphone users. <strong>As</strong> improved multimediafunctionality and capabilities are added tosuccessive generations of mobile devices,<strong>the</strong> stress on existing 3G networks is only setto increase. According to a recent statementfrom Verizon Wireless, <strong>the</strong>ir recently launchedDroid X phones use approximately five times<strong>the</strong> data volume of any o<strong>the</strong>r device including<strong>the</strong> iPhone. Proof that when consumers havea device that works well for accessing <strong>the</strong>Internet, watching video and o<strong>the</strong>r onlineactivities such as <strong>so</strong>cial networking, <strong>the</strong>y willuse it to full effect.Operators are at <strong>the</strong> forefront of this revolutionin mobile data and are struggling withcompeting forces that will determine <strong>the</strong> longtermsustainability of <strong>the</strong>ir business models.Mobile broadband is <strong>the</strong>ir main growth enginetoday, especially in mature markets. But assubscribers flock to mobile broadband servicesand impose increasing bandwidth demands on<strong>the</strong> network, costs are escalating much fasterthan revenues. Operators are faced with a toughchoice: continue to add capacity to alleviatecongestion and meet customer experienceexpectations, or, pay dearly later to combat highcustomer churn and market share erosion.The impact of this increased data usage - morecongested networks - is already starting to showand will have an effect upon users everywhere.Caps on 3G data allowances have already beenintroduced by Sprint and more recently AT&T,Verizon has already hinted at data caps for <strong>the</strong>irLTE network, and o<strong>the</strong>r operators are expectedto roll out similar schemes. Unlimited 3G datadownload packages may <strong>so</strong>on be relegated tohistory. <strong>As</strong> capacity and performance pressuresbuild on <strong>the</strong> network, business and consumerusers who rely on mobile broadband arelikely to experience fur<strong>the</strong>r restrictions on use,inconsistent Internet connections and slowdownloads or delayed interaction.The mobile broadband dilemma and RANcongestionNowhere is <strong>the</strong> mobile broadband dilemmamore challenging, or congestion more acute,than in <strong>the</strong> cost-sensitive Radio Access Network(RAN), which provides connectivity between<strong>the</strong> radio base stations (Node Bs in HighSpeed Packet Access (HSPA) networks) andRadio Network Controller (RNC) hub sitesaggregating and processing mobile traffic.<strong>As</strong> Internet video drives exponential trafficgrowth and rapidly changing traffic patternscomplicate network dimensioning rules, mobileoperators face a number of critical networkissues in <strong>the</strong> RAN that impact both capital andoperational expenses. These include:• <strong>the</strong> need to rapidly upgrade backhaulconnection speeds to avoid trafficbottlenecks;• <strong>the</strong> timing of network equipmentupgrades and investment;• <strong>the</strong> fact that capacity planning isbecoming increasingly reactive;• <strong>the</strong> impact of poor network performanceon customer satisfaction and churn.The ideal <strong>so</strong>lution for this high-cost portion of<strong>the</strong> network must address <strong>the</strong> service deliveryeconomics while adapting to location, networkloads, and <strong>the</strong> traffic mix.Backhaul capacity upgrades: is this enough?One clear choice for mobile operators to<strong>so</strong>lve <strong>the</strong> congestion crunch in this portion of34 • North America 2010
Smoothing Mobile mobile payment broadband Mobile systemsaccess peaks<strong>the</strong> network is to invest in upgrading to faster(higher bandwidth) backhaul connections.The viable options tend to fall into threemajor categories:• Add capacity to <strong>the</strong> existing backhaulnetwork - this is often <strong>the</strong> quickest andeasiest to accomplish operationally.However, it can al<strong>so</strong> be <strong>the</strong> mostexpensive as it <strong>does</strong> not fundamentallychange <strong>the</strong> economics. An example ofthis approach is to add additional T1/EIleased lines or TDM microwave capacityto a cell site.• Upgrade to packet-based backhaul - thisapproach requires significant planning,time, and capital investment. However,it can provide a significant leap incapacity while lowering <strong>the</strong> recurringoperational cost per Mbps of incrementalbandwidth. Examples of this approachinclude upgrading from T1/E1 coppercircuits to fiber fed, Carrier E<strong>the</strong>rnetservices or upgrading from TDM-basedmicrowave to Packet/E<strong>the</strong>rnet-basedmicrowave. Both of <strong>the</strong>se approachesrequire major network, equipment, rightof-wayagreements, spectrum leasing,and operational upgrades that can incursignificant time and cost.• Offload and divert certain backhaultraffic - this approach offloads mobiledata traffic to lower cost transmissionalternatives. Examples of this approachinclude xDSL offload at <strong>the</strong> basestation and <strong>the</strong> use of femtocells andWi-Fi hotspots that can divert <strong>the</strong> dataoff <strong>the</strong> macro network onto cheaperfixed broadband connections. Theseapproaches can be effective but haveeconomic, operational, spectrumplanning, policy and SLA implications.While <strong>the</strong>se options can be effective in anumber of cases, <strong>the</strong>y are typically expensive,reactive, and do not fundamentally <strong>so</strong>lve <strong>the</strong>underlying challenges: exponential growthof mobile data services; <strong>the</strong> cost and time todeploy new transport; and <strong>the</strong> unpredictablenature of when and where congestion will occur.The operational impact of high demandmobile broadband peaksIn <strong>the</strong> course of conducting traffic studieswith a variety of network operators around<strong>the</strong> world, Sycamore has accumulated a rangeof interesting data that sheds light on <strong>the</strong>evolving characteristics of mobile broadbandtraffic and <strong>the</strong> underlying impact of differentcontent types (mix) and usage patterns. Typicalcontent types include video, images, text, ando<strong>the</strong>r binary data that are transported through avariety of protocols between users and servers.The increase in smartphone usage has driven adramatic increase in <strong>the</strong> number of data sessionsin mobile networks, with greater than ten times<strong>the</strong> normal number of session attempts relativeto voice-only handsets.While <strong>the</strong>se trends have clearly increasedbaseline traffic, what is often overlooked is <strong>the</strong>network impact of high demand traffic peaks,which have increased significantly. Most of<strong>the</strong> congestion issues in <strong>the</strong> RAN, in fact, tendto be short-term or transitory, such as spikesin traffic demand during peak hour periods orunpredictable bandwidth surges caused by flashevents. High demand consumption peaks caneasily be four to five times <strong>the</strong> baseline traffic,and are most frequently driven by applicationssuch as Internet video and bandwidth-intensivedownloads of <strong>so</strong>ftware updates.It <strong>becomes</strong> cost prohibitive, not to mentionproblematic, for a mobile operator to size <strong>the</strong>irnetwork (i.e., add capacity) to accommodatethis type of peak traffic since it is not easy topredict where or when those peaks will occur.And while operators will continue to utilizebandwidth expansion options to address point<strong>so</strong>f congestion in <strong>the</strong>ir network, this alone won’t<strong>so</strong>lve <strong>the</strong> quality of service issues that highdemand consumption peaks cause. For <strong>the</strong>serea<strong>so</strong>ns <strong>the</strong>re is growing interest in technologiesthat enable operators to reduce <strong>the</strong> traffic loadstraversing <strong>the</strong> backhaul and adapt in real-time -like a shock ab<strong>so</strong>rber - to <strong>the</strong> dynamics of peaktraffic bursts.Targeting RAN congestion with adaptivecontent optimizationAdaptive content optimization representsa different approach to helping operatorsmanage <strong>the</strong> growth of mobile broadbandtraffic and its attendant operational expense in<strong>the</strong> cost-sensitive RAN. Unlike mobile corebasedoptimization schemes and proxy webcaches, which primarily address upstream andinterconnection bandwidth to <strong>the</strong> Internet,adaptive content optimization can dramaticallylower capacity requirements specificallybetween <strong>the</strong> Node B and RNC. In this part of<strong>the</strong> network, even modest bandwidth (cost)savings can have a disproportionately positiveimpact on profitability.Adaptive content optimization reduces<strong>the</strong> traffic volume traversing <strong>the</strong> RAN byexamining user content flows (including alltraffic types - video, images, text, P2P) andapplying advanced lossless data optimizationtechniques and adaptive learning algorithmsin real time. This approach aims to reducepeak bandwidth requirements in <strong>the</strong> backhaulnetwork and improve subscribers’ quality ofexperience, providing operators with a way tore-balance <strong>the</strong>ir service delivery economics inline with revenue growth - not traffic growth.Adaptive content optimization al<strong>so</strong> provide<strong>so</strong>perators with three important ways tooptimize current network investments whileeasing <strong>the</strong> transition to <strong>the</strong> next generationof infrastructure. First, it addresses <strong>the</strong> maindrivers of mobile broadband: IP video andInternet content - including P2P. Second, it isarchitected for all-IP HSPA radio networks, <strong>so</strong> itis a natural fit with LTE. Third, <strong>the</strong> technologyis scalable to support LTE as a network overlay(<strong>the</strong> way most HSPA operators will deployLTE), <strong>so</strong> existing HSPA sites can be efficientlycombined with LTE radio deployments as <strong>the</strong>yare rolled out.ConclusionMobile operators across <strong>the</strong> globe areexperiencing unprecedented uptake formobile broadband services but strugglingwith <strong>the</strong> resulting network congestion andincreasing costs that come with this success.For US mobile operators in particular, given<strong>the</strong> large populations <strong>the</strong>y must serve andbroad geographies <strong>the</strong>ir networks must cover,<strong>the</strong> challenge will be to continue upgrading<strong>the</strong>ir networks while delivering a good userexperience and managing service delivery costsper MB in line with revenue growth ra<strong>the</strong>r thantraffic growth. Adaptive content optimizationis a compelling way to help operators relievetraffic congestion and <strong>the</strong>reby improve <strong>the</strong> userexperience, delay infrastructure growth, andimprove <strong>the</strong>ir service delivery economics. •North America 2010 • 35