HSDPA : hidden costs and performance issues?

Operators of mobile phone services need to be cautious about upgrading to High-Speed Downlink Packet Access technology. They will need to do more than upgrade the software in base stations to improve the experience for users, says SYLVAIN FABRE...

Some operators of mobile phone services believe that third-generation (3G) technologies have been an underperforming investment so far. So they demand clear justification for yet another technology upgrade.

Vendors may have scaled back their hype about the benefits of High-Speed Downlink Packet Access (HSDPA) and are now giving out more realistic messages, particularly about the speeds it delivers. But operators still need clarification about how deploying HSDPA will affect their networks. It will not be just a software upgrade. And it will not involve spending what was going to be spent anyway later on routine software upgrades to the Third Generation Partnership Project's (3GPP's) Release 99 (R99) networks.

Operators need to carefully examine the need for HSDPA in light of the true end-toend costs of deploying and delivering it. They need to confirm whether HSDPA really is the least expensive way to deliver mobile broadband. They will also need to set the right level of expectations among end users in terms of realistic data rates. In December 2005, Cingular, the largest cellular operator in the United States, announced the launch of HSDPA services in 16 cities, becoming the first cellular operator in the world to offer commercial HSDPA services in more than one market. About 50 other operators have either expressed an interest in HSDPA or are already trialing it. HSDPA is here, and many operators are planning for this next network upgrade. So what should they expect in terms of performance improvements? And what will this really cost?

From Hype to Reality
Data rates experienced by end users will not be as high as communicated, and will vary greatly.Vendors have presented HSDPA as an evolution from WCDMA. They say it will improve end-user experience with a theoretical downlink peak data rates of up to 14.4 Mbps; that it will require minimal changes to existing networks; and that it will require only software upgrades in the existing WCDMA base stations.

However, vendors exaggerated the actual speeds that can be achieved outside lab conditions; and operators have indicated that implementation is not always a simple software upgrade - even for offerings from leading vendors.

Basically, the data rates that will be experienced by end users are much lower than the advertised peak downlink data rate available on the radio channel. In the Universal Mobile Telecommunications System (UMTS), the maximum data rate on the physical channel is about 2.3 Mbps. However, typical data rates for end users are only 384 Kbps. In HSDPA, the maximum theoretical data rate is 14.4 Mbps. However, what users will experience on their terminals may be one order of magnitude lower.

The User Experience
In terms of the user experience, the lower latency of HSDPA does provide a better browsing experience, while higher data rates do provide faster downloads in the best cases. But the quality of service (QoS) in HSDPA is provided as a best effort. The technology's higher data rates are traded off against the QoS inherent in the base technology of WCDMA. As a result, performance for the end user will vary greatly depending on radio conditions, network configuration and the number of concurrent users.

Furthermore, HSDPA focuses on downlink improvements. It is asymmetrical and uploads will not see much improvement over UMTS. Knowledge workers, who rely on more symmetrical traffic patterns, are among the biggest candidates for this technology. It their needs aren't met, they will likely be disappointed and could choose to employ other technologies instead.

To guarantee stable speeds for end users, a dedicated carrier frequency should be used for HSDPA - but this is costly, compared to sharing a common frequency with R99 traffic. For individual users the actual data rates will depend on a combination of three factors: device capabilities, network configuration and user situation.

To guarantee stable speeds for end users, a dedicated carrier frequency should be used for HSDPA - but this is costly, compared to sharing a common frequency with R99 traffic. For individual users the actual data rates will depend on a combination of three factors: device capabilities, network configuration and user situation.

For example, the maximum theoretical downlink speed available to the user is partly defined by the device's processing power. Performance will degrade as users get further from the antenna and closer to the cell boundaries. Also, a user's travel speed will affect performance because fast-moving user equipment is not the focus of HSDPA although the standard specifies it (in 3GPP document TR 25.858). For faster situations, like a high-speed train traveling at approximately 300 kph, performance is still uncertain.

In addition, there is no soft handover support in HSDPA (contrary to R99). So data calls may drop at cell change, which could affect use in some cases, such as downloading music. So it is important that marketing messages focus on the most beneficial use cases for HSDPA, which will be around fixed and semi-mobile broadband usage, rather than mobility.

Impact on the Network
Vendors present HSDPA as a simple software upgrade on the Node B in base stations. However, feedback from operators indicates that, even for major vendors, this may not be the case - especially where they have installed older software releases for the UMTS R99. So, when planning for HSDPA the potential network impacts must be considered, from the perspectives of functionality, capacity and cost. The extent of the network impact will vary but, as a minimum, the Node B will need a software upgrade.

Recommendations
Vendors need to set realistic expectations for what end users can expect to experience to ensure success - you can't just leave it to the operators.

Operators should specify what market needs they are addressing (for instance, metro broadband replacement vs. mobile music downloads) then communicate and plan capacity. Validate customer demand and willingness to pay for applications to drive incremental revenue, to justify the incremental network upgrade costs. Some operators are still trying to build the business case for HSDPA. Enterprise needs are unclear beyond e-mail. And corporations are reluctant to pay for instant messaging services yet. Just reducing cost per megabyte transmitted may not justify the upgrade to HSDPA.

Consider delaying HSDPA rollout plans until demand for bandwidth is confirmed, or consider leapfrogging HSDPA and upgrading later directly to HSDPA/HSUPA as an investment with a better net present value (NPV). At launch, communicate realistic and committed data rates to subscribers to protect brand equity. HSDPA will enhance the mobile data experience, but it will not be like a fixed digital subscriber line (DSL). By creating a business environment where not only the operator but also partners can find compelling new applications, the ecosystem will breed success.