The Uplink Bottleneck in Early HSDPA
The introduction of High-Speed Downlink Packet Access (HSDPA) substantially increased download capacity and reduced latency in 3G UMTS networks. Users could browse media-rich websites, stream music, and watch videos more comfortably on mobile devices. HSDPA focused on one direction of traffic, the downlink from the network to the user, so the uplink remained comparatively limited.
While downloading became faster and more responsive, uploading information from the phone back to the network remained near the original UMTS reference rates, with values such as for the initial service profile. Early internet use was dominated by content consumption, while later interactive and user-generated applications required a faster uplink. Examples included:
- User-Generated Content: The rise of social media and photo/video sharing platforms meant users were no longer just downloading; they were uploading high-resolution photos and videos from their phones.
- Interactive Applications: Real-time applications like video conferencing, interactive online gaming, and voice-over-IP (VoIP) calls require a fast and low-latency connection in both directions.
- Cloud Services: The shift towards cloud computing meant users needed to upload large files, sync documents, and back up their devices to remote servers.
The slow uplink became a bottleneck for mobile internet services. To extend the 3.5G evolution with a more balanced data service, 3GPP developed a counterpart to HSDPA: High-Speed Uplink Packet Access, or HSUPA.
What is HSUPA? Improving the Uplink
High-Speed Uplink Packet Access (HSUPA) is an enhancement to the UMTS (3G) standard, specified in 3GPP Release 6. It applies fast scheduling, retransmission, and power-control procedures to improve uplink data speed and latency. Together, HSDPA and HSUPA form the technology family known as .
The uplink in a mobile network requires coordination of transmissions from many independent, battery-powered devices distributed throughout a cell. The network must manage these transmissions to avoid the and control the overall interference level. HSUPA therefore introduced technologies related to those in HSDPA, adapted for the constraints of the uplink direction.
The core technologies of HSUPA
HSUPA's performance gains rely on technologies related to its downlink counterpart, HSDPA. These include fast scheduling, rapid error correction, and configurable short transmission intervals, all optimized for uplink traffic.
1. Fast uplink packet scheduling: The request-grant system
Just as with HSDPA, a key innovation of HSUPA was moving the resource scheduler from the distant RNC to the much closer Node B. However, uplink scheduling is fundamentally different from downlink scheduling. In the downlink, the Node B has a complete picture of all the data waiting to be sent to all users and can opportunistically choose who to transmit to. In the uplink, the Node B does not know how much data each user has in their buffer waiting to be sent.
HSUPA solves this with a request-grant mechanism:
- The Request (Scheduling Information): When a User Equipment (UE) has data to send, it cannot just start transmitting. Instead, it sends a request to the Node B on a new dedicated control channel. This request, known as , tells the scheduler in the Node B two critical things: how much data is in the UE's buffer and how much power the UE has available for transmission (its power headroom).
- The Decision (Node B Scheduler):The Node B's scheduler receives requests from many UEs simultaneously. Its job is to manage the total amount of uplink interference, which is a shared resource for the entire cell. Based on the requests it receives, network policies, and the overall interference level, the scheduler determines which UE transmissions receive resources in upcoming scheduling intervals and what maximum E-TFC or power level is allowed.
- The Grant:The scheduler communicates its decision to the UE by sending a "grant" message on new downlink control channels. This grant tells the UE which transmission level or transport-format combination it may use within the current interference limits.
2. Hybrid ARQ (HARQ) with soft combining
HSUPA employs a HARQ mechanism related to that used by HSDPA, operating in the reverse direction. It provides fast retransmission and correction of transmission errors on the uplink.
- Uplink Transmission: The UE sends its data packet to the Node B in its assigned time slot.
- Fast Feedback from Node B: The Node B hardware immediately tries to decode the received transport block. It then sends an ACK or NACK for the associated HARQ process to the UE on the downlink feedback channel.
- Rapid Retransmission from UE: If the UE receives a NACK, it retransmits the associated data, often within a few milliseconds. As in HSDPA, soft combining lets the Node B store information from the failed transmission and combine it with the retransmission, increasing the chance of successful decoding. This hardware-managed error-correction loop reduces uplink latency.
3. Shorter transmission time interval (TTI)
HSUPA supports both and 10 ms E-DCH TTI values in FDD. A 10 ms TTI is mandatory for FDD UEs, while 2 ms support is required only for specified UE categories and is selected through Layer 3 signalling. The shorter interval can make request-grant exchanges and HARQ retransmissions more frequent than in the original 10 ms configuration, reducing uplink latency for interactive applications such as video calls and online gaming. The selected TTI is therefore a UE capability and configuration detail. The E-DCH category also affects the transport-block size available in each interval.
The new HSUPA channel structure: Enabling high-speed uplinks
To implement the request-grant mechanism and HARQ feedback, HSUPA introduced a completely new set of physical channels. This new transport channel for uplink data is collectively known as the . The E-DCH consists of several associated physical channels, some on the uplink and some on the downlink.
- Enhanced Dedicated Physical Data Channel (E-DPDCH): This is the main data-bearing channel. When a UE receives a grant to transmit, it sends its actual data packets (e.g., a photo upload, video stream) on one or more E-DPDCHs. A UE can be assigned multiple E-DPDCHs simultaneously to achieve higher peak data rates.
- Enhanced Dedicated Physical Control Channel (E-DPCCH):This channel runs in parallel with the E-DPDCH and carries the critical control information that the Node B scheduler needs from the UE. This includes the Scheduling Information (buffer status, power headroom) that forms the "request" part of the request-grant cycle.
Interestingly, to make the uplink faster, several new downlink channels were needed to provide control and feedback from the Node B back to the UE.
- E-DCH Absolute Grant Channel (E-AGCH):This is a shared downlink channel used by the Node B scheduler to send an "absolute grant" to a UE. This message sets an absolute grant limit within which the UE selects an allowed E-TFC and power level. It is typically used to give a new UE permission to start transmitting or to update an existing UE's transmission level.
- E-DCH Relative Grant Channel (E-RGCH):This dedicated downlink channel provides very fast, incremental adjustments to a UE's data rate. The Node B can send a single "UP" command to allow the UE to increase its rate slightly, a "DOWN" command to force it to decrease, or a "HOLD" command. This allows for very fine-grained and rapid control over the overall cell interference.
- E-DCH HARQ Indicator Channel (E-HICH): This dedicated downlink channel carries HARQ feedback. The Node B transmits an ACK or NACK for the associated transport block and HARQ process, allowing the UE to determine whether retransmission is needed. One E-HICH transmission can carry feedback for multiple UEs that require acknowledgement in the frame.