Packet-switched 2.5G and 3G network deployments will enable operators to layer data- and content-centric services onto existing voice capabilities. But speed alone is not sufficient to ensure the success of streaming media and other next-generation services.
With wireless mobile networks, operators will need the ability to provide highly personalized services based on subscriber profiles, affiliation and location, the characteristics of the data content, and its value to the subscriber or even a third party.
Because voice will continue to be the killer app, mobility, service granularity and personalization will have to be the killer attributes that will drive users to wireless data services, according to Yankee Group program manager Phil Marshall, who has written several white papers on the topic of mobile IP. “Operators must draw together their business and networks in order to find value in advanced data services; therefore, they must integrate IP service equipment and subscriber awareness,” he says.
Not All Bits Are Created Equal
Before they can incorporate mobile IP services into devices, operators must first assign value to data services, as charging per megabyte will eventually erode high-volume services like SMS. Operators run the risk of commoditizing their networks if they become mere transport mechanisms for third parties moving wireless data to subscribers. To become mere providers of access would be detrimental, as many operators have committed huge amounts of capital to licenses for next-generation networks. To succeed, operators need granular subscriber and content information for segmentation, so they can derive value from content, and consequently control the edge of their IP networks.
“In order to maintain control of subscriber services, carriers must begin to enhance their IP backbone capabilities for wireless data services now, rather than wait for 3G,” says Marshall.
Of course, overlaying an IP backbone on top of an existing circuit-switched network is not enough to integrate service intelligence for real-time interrogation and aggregation of packet data in mobile IP networks.
Such interrogation and manipulation of packets must be handled by mediation. However, traditional mediation was born out of the fact that most routers were dumb, since they lacked any detailed view of networks. With mobile IP applications such as VoIP, VoD and videoconferencing, there will be a need for smarter mediation—mediation that offers granular billing and revenue information for transaction-oriented IP services that involve myriad variables for traffic volume, QoS and content value, as well as complex relationship management.
Next-Generation Mediation
There’s no question that conventional billing, rating and mediation systems struggle to participate in online charging, as they do not have a real-time orientation. Systems built around circuit-switched networks focus on call duration information derived from CDRs, issued after events have occurred. Consequently, there’s a lot of data going up layers of protocols, whether WAP, TCP/IP services, or others.
As a result, we are starting to see a slew of premediation companies that mediate among traditional mediation, billing and intelligent network (IN) platforms. They target the mobile Internet or edge development for a more data- and consumer-aware network.
Companies like Watercove, Megisto, Tahoe Networks, and ProQuent Systems (formerly known as Avian Communications) are being flooded with venture capital to create edge devices and routers that are IP- and user-aware, so that operators can have a network-centric approach with an eye on the business aspects of delivering wireless data services. They all claim to be able to give operators control and visibility into their networks.
The mission of each of these companies is to build next-generation, intelligent, service-aware systems that integrate with carrier-class switching and routing platforms for CDMA, GSM and UMTS networks.
Simply stated, premediation systems sit between the switch and the mediation systems in order to kill off superfluous IP data, thus enabling data service details to be processed in real time—for real-time charging.
These boxes and next-generation routers are designed for mobile networks, thus being able to handle packets of data in a more efficient manner than conventional systems.
Rather than rely on network element managers to build equipment, solutions from premediation companies sit in the flow of packets, taking a more active role in shaping traffic as well as policy management.
A Closer Look
Several companies have started making noise in the premediation space, and while they take different approaches to the concept they are all focused on giving service providers a clearer view of their network traffic.
Watercove
“Where traditional mediation meant looking at charging record streams, premediation means being in that stream, actually affecting its flow,” according to Mark Tubinis, CTO of Watercove Networks, which offers a packet data system that integrates mobility and service intelligence with metered transport—the ability to measure and control data transport in real time. Watercove provides the transport path for enforcing QoS events, such as charging on the fly, so operators can work toward real-time charging and flow-based visibility.
“Ultimately, wireless carriers have to tie the mechanisms into the transport that enables real-time charging functions, so that they can measure not only volume and time for a session, but levels of detail about the flows within a session, and consequent time- and volume-based measurements to those as well,” explains Tubinis.
For example, if a customer wants to boost QoS for higher-speed transmission over 30 minutes for uploading files during an impending meeting, or if a user wants a multimedia message, the request needs to go to a multimedia message server, and from there the premediation system follows the logical flow of information. By following the request and response, it is understood whether the request is still outstanding, and whether a session or flow number has to be maintained. As the system looks for a response coming in from the other end, it coordinates it with the initial request through a transaction ID the system stores for every flow. The system then looks for responses with transaction IDs to match them up.
“It’s the premediation system that tells the service layer how many packets of data were consumed and for what period of time. It then alerts the rating and billing engines that a charge should be assigned to that consumer,” explains Tubinis. He notes that operators should seek out premediation systems that can scale to measure the volume, time and event charging situations, and apply them across a broad number of sessions and flows within those sessions.
That type of end-to-end QoS capability will begin to pick up momentum in the United States in the next year, Tubinis believes: “Once operators get through the complications of the first roll-out, it will take off,” he says.
In the meantime, Watercove is forging relationships with an array of billing, mediation, IN equipment and radio access network vendors. “The more cooperation there is among all the elements,” Tubinis says, “the easier it will be behind the scenes once operators decide to roll out next-gen wireless data services.”
One of the challenges Watercove is working on is getting the incumbent network providers to participate in the radio access network side. “They have to make radio access controls visible to the rest of the network,” Tubinis says, “before they go on to PCF, PDSNs, SGSNs, and GGSNs—core networking components in MPLS.”
Megisto
Megisto is in the same broad market as Watercove, as its Mobile Services Delivery System (MSDS) is a subscriber- and service-aware core infrastructure designed with a focus on high-value, personalized mobile data services.
“Operators have little subscriber knowledge as they build new packet core networks,” says Carol Politi, vice president of marketing at Megisto, which hopes to help operators figure out how to handle packets and then how to charge for them.
While she acknowledges that operators are getting better at managing subscribers as they move around the network, the concept of treating packets individually is still foreign to them. “It requires they link IP service equipment and subscriber-aware systems,” she says.
To achieve that, operators must have infrastructure that interfaces with charging systems and maintains multiple charging records simultaneously, as is done in phone networks today. “Just as voice-centric environments enable operators to view a history of calls so that it is clear what was charged and for what transactions, data usage or per-transaction charging will be necessary if carriers are to make money from their networks,” Politi says.
If operators are to recoup money spent on building out networks, she says, they must have charging facilities that can handle reverse charges and incentives that will be beneficial to subscribers. “The systems must also be balance-aware, capable of handling the accounting for pre- or post-pay. You don’t want to just cut off a subscriber if you can recharge their account by passing them onto an operator,” says Politi. That approach will require premediation systems to interface with existing rating and billing systems so that they can access customer charging policies in real time.
Operators, she says, must also link into service management systems, so they can extract profiles from voice-centric infrastructure and put them into mediation systems that will charge for the service.
“Operators need to realize that end-to-end billing problems will not be resolved with a single piece of infrastructure, as back-ends are complicated,” she emphasizes. “No existing or up-and-coming vendors can solve that end-to-end.” Instead, Politi suggests that operators find preintegration companies that elegantly integrate with what exists.
She warns against focusing solely on driving subscribers to the network. “Simultaneously, operators should look at whether their infrastructure supports that; they should realize how critical it is that the infrastructure scale if services take off,” she says. “If it’s anything like the SMS roll-out, the uptake was initially slow, but then it exploded into an exponential increase.”
ProQuent
ProQuent Systems’ IP Mobile Services Switching Point offers a set of network elements that enables wireless network operators to collect information or trigger services based on the content and the state of a mobile data user’s IP session.
The product connects to existing wireless network elements (CDMA, GSM/GPRS and UMTS) via standard interfaces, offering what the company calls “intelligent mobile infrastructure” by placing the box directly into the traffic flow.
The technical difference between ProQuent and its competitors, according to Jules Meunier, president and CEO, is that most are based on GGSN (Gateway GPRS Serving Node) networks.
“Services that are GGSN-dependent are complicated, because you have to have all suppliers providing the same services, which requires that operators synchronize vendors to deploy services across the network,” he says. While ProQuent does have network elements that are located in the GGSN to handle mobility management, security and tunnel terminations, operators insert NEs between GGSN and the Internet, thus enabling real-time CDR capabilities and rating, which leads to real-time billing for data and voice.
In order to really know what users are doing, ProQuent’s system intercepts the data flow at the text stream level, not just the URL level.
“We take the extra IP information out, so if an operator has an advice-of-charge application, data flows today dictate that all bits about the user and URLs go to that application server. But we parse through all the data and automatically forward less traffic to that server,” says Meunier.
Advice-of-charge is one area where ProQuent is focusing, because dynamic pricing is necessary to apply special discounts and offers in real time. “We simplify things for the application server architecture, so it is more scalable,” Meunier says.
“Third-party billing and sponsored services, such as Freephone, or 800 voice services, where content providers pay wireless operators for transactions with the end user, will soon emerge,” he predicts. He also believes sponsored services will be tightly linked to prepaid accounts where wireless subscribers’ accounts will not be decremented for the transaction or the Web session. ProQuent currently is honing its own applications for IP accounting so that granular, real-time billing won’t be left up to conventional mediation, rating and billing systems.
Tahoe
Tahoe Networks’ Mobile Internet Edge (MIE) is a wireless infrastructure designed to let carriers manage traffic as well as subscribers by class and policy. “That requires a revenue-sharing paradigm between operators and content providers. It’s not a capability you will find in a Sun server parked next to a plain-vanilla router,” says Alan Cohen, vice president of marketing and product management. “Vendors are still building equipment that way, but what’s needed is granular software so that the operator can know, for example, this customer is using an air card and that a bill should be sent to her for downloading a horoscope from Yahoo into her personal account, at 30 cents per horoscope.”
For more comprehensive revenue sharing, Tahoe partners with Xacct to deliver in-depth billing and revenue information in policy management. “They need us to produce smarter information so they can support more granular billing and revenue information, rather than just air time and bits information,” says Cohen. “Instead of Xacct having to reconstruct events, we provide infrastructure so the operator can get paid for the actual IM message, for example, rather than just the data download and connectivity.”
Instead of just taking out unnecessary data from routers —such as headers and addresses that are not germane—and parse up to billing, Tahoe’s MIE can be a captive portal, so Tahoe routing and mediation can handle services like Reuters news feeds or VPN access.
Tahoe eliminates the ping proxies, for example, that send messages out every 15 seconds or so about IM traffic. “Most of it is inane signaling traffic, so we can maintain that and eliminate that extra data, freeing up spectrum,” Cohen explains. MIE also authenticates users, and tells messaging servers when users are connected through Layer 3 systems (as opposed to WAP gateways).
Traditional Mediation Will Complement Premediation
Although mediation for IP mobile services will be different from traditional mediation, premediation systems will complement traditional mediation, not replace it.
Narus
“It’s the premediation companies that will provide the raw data in a more detailed level than did classic NEs, so they will offer a preprocessing capability not possible before,” says Sue Forbes, vice president of marketing and business development. “That complements us in that our power is in central aggregation, filtering and correlation, looking across multiple networks that go into billing systems for these services.”
While the premediation companies will pull information in unit by unit, mediation companies will put the information in the final correlated record, which is how billing usually likes to see the information. Forbes notes that Narus already has implementations on content-based systems with KDDI in Japan, which will test Narus to make sure its systems can scale to a half-billion records a day. “With their content-based billing,” she says, “we are the mediation component providing the whole for that detailed information per URL, per subscriber.”
Xacct
“The box has to be user- and service-aware, which is where the premediation companies come in. They know more than just the IP address; they know the subscriber, too,” says Anil Uberoi, senior vice president of marketing and business development at Xacct Technologies. “Where old mediation boxes provided just a connection, premediation provides the intelligence for routing of packets and exposure to granular data. That is the value of premediation, as it provides a policy-based switching of packets, based on more detailed data based on the flows.”
Because many operators already have existing prepaid engines, or some pricing server or pricing policy engine somewhere, it all must be synchronized if operators are to present real-time offers to online mobile subscribers.
Xacct takes information out of the premediation boxes and interfaces with balance management, price lists, time-of-day and geography data, to decide how it impacts the special offer on the back end. “All policies live in the back office in the balance manager or IN systems. So we must go beyond just taking CDRs and pumping them into billing systems; now we must plug the hole between the hardware and back-end systems,” explains Uberoi.
Xacct has launched its advice-of-charge solution, which will give carriers dynamic pricing based on user profiles. “That is a hot button in mobile now,” says Uberoi. “From the time a subscriber picks up the phone, everything has to happen in real time. Where traditional billing didn’t need real-time mediation, reverse charging and now advice-for-charge applications will require real-time mediation.”
As operators move to SMS and MMS, carriers may want convergent mediation platforms. “They will not want two systems for pre- or postpaid customers,” Uberoi says. “They will want one engine for both, as many customers, particularly abroad, have two accounts—personal accounts, which are prepaid, and business accounts, which are postpaid.”
Openet
“Operators will always need a hardened mediation platform,” says Joe Hogan, CTO and founder of Openet, whose FusionWorks platform is designed to deliver high volumes of traffic on the order of thousands of events per second. “IP services generated by GPRS traffic will contain many different elements, as well as content services delivered by various gaming servers, weather servers, traffic servers, and so on—each with different accounting protocols and models,” he says. As a result, operators will see IP routers and authentication servers from multiple companies. “There is no standard set of equipment,” Hogan says, “so the challenge for mediation is to rapidly collect all the necessary data from new data elements.”
Openet is banking on its heavy engineering focus as a differentiator. “While our competitors spend millions of dollars on marketing and educating about charging for data services,” Hogan says, “we concentrate on the engineering of a scalable performance system for IP mediation.”
He explains that in a premediation or active mediation environment, where the mediation system receives an indication from NEs that it will deliver traffic to a subscriber, it requires clearance from the existing mediation system.
Openet’s system uses rules-based engines. “They correlate inbound streams of data—each giving you several pieces of the overall puzzle, so you can rapidly assemble a picture of what the subscriber is all about,” Hogan says. IP routers, GPRS equipment and the services themselves have to communicate before operators can decide how to bill for services. “We have built our correlation strategy around rules so we can read multifarious inbound feeds,” he says. “Otherwise, you have to rewrite and rebuild each time.”
If You Build It,Will They Come?
While the incumbent mediation vendors and the emerging premediation companies are bolstering their systems to support mobile IP needs, thus far only the most aggressive carriers have MPLS-enabled networks. Most can be found in Asia and Europe, as companies like NTT DoCoMo, KDDI, Orange and BT look to mobile IP as another source of revenue.
In the meantime, premediation companies are not sitting on the sidelines. Most have been aggressively partnering with manufacturers to make marked improvements in equipment, as evidenced by a slew of hybrid processors emerging from the top chip manufacturers—hybrids combining network processors, which excel at handling packet manipulation, and processors (like power PCs and Pentiums), which excel at complex accounting and charging. That strategy will help marry the network to data- and subscriber-aware data.