Test & measurement
Nick Hallam-Baker looks at some of the potential issues that can arise with the introduction of enhanced air interfaces.
ellular networks based upon High-Speed Downlink Packet Access (HSDPA) technology marks a clear shift toward the support of packet data services. HSDPA networks offer higher data rates and reduced packet latencies to provide new services comparable with home broadband experience. However, HSPDA addresses only half the story, and there are benefits to be gained by enhancing the uplink.
You no doubt have seen headlines extolling the 14 Mb/s downlink data rate offered by HSDPA. This is a very important feature of the technology and provides services comparable to those offered by WiFi hotspots plus the advantage of improved mobility, security, and ease of use of a cellular network. HSDPA marks a major step in the evolution of cellular networks to a system specifically optimized for downlink packet data. And, as we will see, HSUPA extends this packet data enhancement to the uplink.
Evolution of 3G networks
With the introduction of HSDPA, significant aspects of the management of the air interface have been moved from the radio network controller (RNC) down into the base station or Node B in 3G terminology. Moving this functionality closer to the air interface means that the 3G system can react more quickly to changes in the quality of the wireless link between the user and the Node B – and to the user’s data requirements. In turn, this efficient and reactive control allows for higher data throughput and greater cell capacity.
The higher data rates associated with HSDPA are achieved through the use of advanced channel coding and modulation techniques with the backup of fast retransmission in case things go wrong. To take advantage of these features, the Node B receives information from each user about the quality of the downlink channel.
In times of very good channel quality, the Node B can transmit very high data rates with little error protection from the channel-coding algorithms. However, as the channel quality decreases, the effective user data rate is reduced since the Node B must send more robust information with greater error protection. Depending upon the environment, the user’s channel quality can change rapidly, and the Node B must react quickly if it is to maintain maximum efficiency of the air interface.
Another major advantage of HSDPA was the introduction of a shortened packet duration or time transmission interval (TTI). Reducing the minimum TTI from 10 ms in a Release 99 system to 2 ms for HSDPA provides several advantages. First, it allows the Node B to react more quickly to changes in the channel quality experienced by each user. But perhaps more importantly, it reduces the latency associated with the transmission of each packet. The packet latency represents the delay between the data packet transmission and its successful reception and decode.
For many services, such as voice over IP (VoIP) or interactive gaming, excessive packet latency rapidly degrades the quality of service. When coupled with hybrid automatic retransmission request (ARQ), allowing fast retransmission of erroneous packets, the reduced latency through the network enables the cellular operator to offer a much wider range of services.
The need for HSUPA
Given the improvements in efficiency and services provided by HSDPA, the next step would be to apply similar concepts to the uplink. High-Speed Uplink Packet Access (HSUPA), sometimes termed Enhanced Uplink, represents the latest 3GPP Release 6 technology and aims to provide optimized packet data support in the uplink.
However, some industry commentators have said it is the downlink that is critical, with uplink usage being much less important. Consequently, is there any need to introduce this technology? While it is true that many broadband services currently are dominated by downlink data transfer, it is important to recognize the need for efficient packet data support on uplink.
Services such as multimedia messaging are already popular and likely to grow as high-quality cameras become standard in most handsets. Additionally, it is likely that the use of symmetric services such as voice and VoIP will start to increase, both offering more efficient use of bandwidth than their circuit-switched counterparts.
Finally, in recent years, TCP/IP has become the ubiquitous transport mechanism for packet data, and the physical (PHY) and medium access control (MAC) layer aspects of HSDPA are ideally suited to supporting this higher-level protocol. However, downlink TCP/IP services require corresponding packet acknowledgements to be sent on the uplink and, with a high rate downlink, this alone can generate a significant load on the uplink.
If acknowledgements on the uplink are lost or delayed, TCP/IP may also adapt to its perceived link quality by reducing the downlink data rate. Mixing and matching HSDPA downlink with the acknowledgements transmitted on conventional Release 99 uplink channels present inefficiencies in terms of uplink bandwidth and latency.
HSUPA features
Many of the features and enhancements of HSUPA can be traced directly back to HSDPA. Additionally, new challenges have been addressed to balance the need for efficient management of the uplink air interface against costly control signaling and implementation complexity.
Like HSDPA, HSUPA offers enhanced data rates, fast packet retransmission mechanisms, and reduced packet latencies. The uplink data rate for HSUPA is increased up to a theoretical maximum of 5.76 Mb/s.
One of the techniques used to achieve this is adaptive channel coding, which adjusts the amount of error correction according to load and channel conditions. Hybrid ARQ (HARQ) packet transmission techniques and the 2-ms TTI also are copied from HSDPA.
The support of the high data rates and the reduced TTI present a number of complexities and challenges to both handset and infrastructure designers. From the handset perspective, the support of the high data rates necessitates many considerations within the radio, baseband and protocol implementations. More powerful DSPs, faster ASICs, and increased memory are required to handle the larger amounts of user data and faster data processing.
The reduced TTI requires that the handset must react more quickly to retransmission requests and control signaling from the network. The high data rates may also require improvements in the quality of the components in the RF transmitter stage to ensure that the encoded data is not corrupted by its own radio.
Implementation of these features will take time and can add significant cost to the manufactured price of a handset. For this reason, a range of six handset capability classes has been defined supporting different data rates and different TTIs. The lower classes accommodate lower data rates, starting from 700 kb/s, and some classes only handle a 10-ms TTI. The first commercial handsets will likely conform to these lower capability classes with the higher rates being rolled out as the technology evolves.
Infrastructure manufacturers face similar pressures when updating their Node Bs to support HSUPA with their requirements for higher quality radio receiver architectures and improved baseband processing. Additionally, cellular operators are likely to pressure their infrastructure suppliers to provide support for the high data rates from the initial release to minimize their own rollout costs and timescales.
Control and scheduling
Perhaps the biggest challenge associated with HSUPA is not the physical transfer of data over the air interface, but its management. The air interface of an HSUPA cell must be carefully controlled to ensure that each user receives the uplink bandwidth required while preventing cell overload due to too many users trying to send data at the same time. This is the task of the scheduling algorithms.
Like HSDPA, the scheduling algorithms for HSUPA are located in the Node B, enabling quick responses to changing channel conditions and user data requirements. In itself, this presents an interesting problem.
For HSDPA, the Node B controls the downlink resources and possesses all of the information required to share those resources among all users. For HSUPA, the situation is more complex since information about uplink data bandwidth requirements resides with each user and little information is available to indicate the uplink channel quality for each user.
To add to the challenge, an inferior HSUPA scheduling algorithm could result in each user generating excessive interference in the cell. This would cause an overload and mean that no one succeeds in getting data over the air interface.
While each Node B manufacturer must define and optimize its own HSUPA scheduling algorithms, the mechanism of communicating the control information between network and handset has been defined by the 3GPP standards. The challenge was how to co-ordinate multiple users, each with their own specific data requirements, while minimizing the control signaling and associated delays. An example of HSUPA uplink and downlink signaling is provided in Figure 1(above, left).
First, the Node B needs to know whether each user has any data that it wishes to transmit. Two methods are defined by the 3GPP standards to address this problem.
Each user may periodically inform the network about the status of its uplink data buffers. This Scheduling Information (SI) includes details about the amount of data waiting to be transmitted and its corresponding priority.
While the SI contains many details, it is complex to encode and presents a comparatively large overhead in signaling data. Consequently, a second method provides just a single bit of information that allows the handset to indicate whether the Node B has allocated sufficient resources for it to transmit its data. Thoughtfully termed the Happy Bit, this information is transmitted every 2 ms and allows the Node B to fine tune its uplink resource allocations.
With a method defined to indicate the amount of data each user wishes to send, the next question is how does the Node B inform the user of its bandwidth allocation? The solution is for the Node B to allocate each user with a transmission grant that indicates the proportion of his transmit power that may be used to send HSUPA data. The grant effectively equates to a maximum data rate at which the user is permitted to transmit until a new grant is issued.
Two mechanisms are provided for signaling this grant: the Absolute Grant and the Relative Grant. As the names suggest, the Absolute Grant specifies a precise value. The Relative Grant simply indicates a single step offset.
Since the Absolute Grant contains more information and a larger downlink signaling overhead, it is likely that it would be used relatively infrequently. Instead, the Relative Grant, requiring a much lower signaling overhead, could be used for fine adjustments to the operation of each user. Further reductions in downlink signaling overhead can be achieved by sending the same grant information to multiple users by sharing the Absolute or Relative Grant channels.
Soft handover
A final twist to the scheduling management is the capability of an HSUPA network to support soft handover. Soft handover, a technique widely associated with CDMA systems, is the process where a user receives data from and transmits data to multiple base stations. This provides particular benefits at the cell edges where signal quality might be poor.
For an HSUPA user, the importance of soft handover has two aspects: power control and resource allocation. First, as a user transmits data, he generates noise-like interference to other users within nearby cells. If a user is near the cell edge, he will have to transmit at a higher power to compensate for being farther away from the Node B, causing greater interference. This effect can be offset by soft handover where the uplink data is received and combined by multiple base stations, reducing handset transmit power and the resulting interference.
The second consideration applies to resource allocation. For example, an HSUPA user in a lightly loaded cell may be given a grant that allows it to transmit at high data rates, maximizing the benefit to that user. However, if the user then moves toward a heavily loaded cell, his transmissions could result in excessive interference. The support of soft handover in HSUPA provides a mechanism for the new cell to directly manage the interference from this new user.
The HSUPA network handles the uplink data rate from each user by means of allocating grants. By controlling the data rates, the network manages the interference generated by each user.
Figure 2 (above) illustrates the network management in a typical soft handover scenario. At any instant in time, each user has a master serving cell. The serving cell is responsible for managing the uplink data requests from each of its users and allocating appropriate grants. In a soft handover scenario, other non-serving cells may be configured to decode the uplink HSUPA data.
In the example in Figure 2, the serving cell is lightly loaded, and the user has been issued a large grant. However, one of the non-serving cells is supporting a higher traffic load. Therefore, the non-serving cell is able to signal a Relative Grant to users to request that they reduce their transmit power.
This mechanism is a key feature of HSUPA since it allows different Node Bs to manage the uplink air interface without involving the RNC. This, in turn, means that the network can react much more quickly to changes in user data requirements and interference scenarios to maintain optimal resource allocation.
HSUPA: slideware or reality?
HSUPA represents a significant enhancement in the evolution of cellular technology. In conjunction with HSDPA, it paves the way for mobile, easy-to-use, and secure broadband services that we currently tend to associate with fixed wireline or short-range wireless systems. What is more, the efficient use of the air interface will provide improved quality and a wider range of enhanced services.
With network trials planned for early 2007 and rollout commencing later that year, HSUPA is a reality. The leading network manufacturers already are at the complex stage of testing and validating HSUPA Node B implementations.
In the future, cellular systems will continue to evolve, and the 3GPP standards groups are already working on their long-term evolution. By considering upgrades and enhancements to the network architecture, new air-interface modulation schemes and advanced multiple RF antenna techniques, cellular capability will continue to increase to meet the expected growth in mobile data services.