Dividing '3' into 2: what the Hutchison-Telstra deal means for users

Telecommunications Journal of Australia, Winter 2005 by Chapman, Paul, White, Lang

This paper questions the impact on end users of the agreement between Hutchison and Telstra to share the spectrum underlying the '3' network, given that congestion in W-CDMA makes quality of service variable. It discusses the technological and regulatory situation and suggests that greater disclosure of relevant details in the agreement along with improved monitoring by the ACA and more choices for end users are all needed.

INTODUCTION

In August 2004 Hutchison and Telstra agreed to share the spectrum underlying Hutchison's '3' network.[1] It's not hard to understand why: on the one hand Hutchison has accumulated significant losses on its 3G in Australia; on the other, Telstra has spent $300 million on 3G spectrum but has put off spending more to rollout its own network using that spectrum. The sharing agreement addresses both problems. It is a way of making use of Hutchison's spectrum more quickly and a way for Telstra to get into 3G more cheaply than either could have done alone. It's probably good for the carriers but is it good for users?

Statements by the carriers at the time of the announcement told us the deal will be good for everyone because it will enhance 3G competition. But wait a minute - weren't we told that competition works because it weeds out the inefficient in a deadly rivalry? If Hutchison cannot make a go of 3G as they had intended, the network should pass cheaply to an operator able to make better decisions. It is hard to see how competition is enhanced when the competitive process has been curtailed by comfortable agreements between rivals. But let's leave that aside: perhaps the good to come from competition has been overstated.

In this paper we want to focus on what seems to be a sleeper in the agreement, an aspect likely to count against the interests of users but one not willingly discussed by the carriers. The nub is this: the W-CDMA technology behind the '3' network will make users interdependent because, as the load on a cell increases, the resources consumed by one user reduces the quality of service (QoS) available to others. Managing that interdependence is hard enough when the network is run by the carrier which owns it. When the network is shared, the complications increase rapidly raising a raft of yet unanswered questions regarding the agreement.

To explain the situation we begin by outlining the W-CDMA technology, describing how code division enhances capacity, how interference creates the interdependency and how that creates management issues. Then we show that '3' will be profitable only if the networks allow loads so high that this interdependence comes into play. The views of carriers and the regulator are then reported and we conclude with some critical comments and suggestions for improvement.

THE TECHNOLOGY BEHIND W-CDMA

Traditional wired telephony and second-generation wireless systems operate by allocating to each user a dedicated channel for the duration of the call. Each user can be assigned a unique frequency channel (as in wired telephony) or unique timeslot (as in GSM) and, provided the system is functioning correctly, there is no interference generated among users. When all available channels are assigned, further requests for access to the network will be denied and the user attempting access will be blocked. In short, blocking is the only congestion effect and, apart from that, users may be thought of as independent.

In CDMA (Code Division Multiple Access), the technology behind 3G networks, the allocation of system capacity is fundamentally different. All users transmit simultaneously on the complete frequency band. Users are separated by allocating a unique digital 'spreading code' to each. This approach greatly increases the efficiency with which the spectrum can be used but, generally, for capacity reasons, these codes do not perfectly separate users, and thus there is always some level of interference between them. In some systems, orthogonal spreading codes are used with the aim of mitigating the interference (called Multiple Access Interference, MAI); however multipath radio propagation effects tend to degrade the orthogonality and thus result in MAI even when orthogonal codes are used.

A schematic of the CDMA process is shown in Figure 1.

This all means that, unlike previous network technologies, CDMA networks do not have a hard limit on the number of users which can be admitted. However, admission comes at a cost to others because in a CDMA network as more users are admitted, the overall interference level experienced by all users will increase. Eventually, some users will experience a drop in their signal-to-interference ratio (SIR) which is so significant as to violate their target QoS. At the discretion of the network operator these users can increase their transmitter power to redress the effects of interference.

However, if some users start to increase power, others will experience increased interference, and some of those might increase their power to attempt to compensate. Eventually, the rise in SIR will be so significant that many users will find the QoS unacceptable. This capacity constraint is called the soft limit of a CDMA network. It is a fundamental property of CDMA networks.[2]