UMTS/HSPA State Transition Problems to be solved with LTE

The way UMTS/HSPA is designed is that the Mobile (UE) is always in IDLE state. If there is some data that needs to be transferred then the UE moves to CELL_DCH. If the amount of data is very less then the UE could move to CELL_FACH state. The UE can also move to CELL_PCH and URA PCH if required but may not necessarily do so if the operator has not configured those states.

The problem in UMTS/HSPA is that these state transitions take quite some time (in mobile terms) and can slow down the browsing experience. Martin has blogged about the state transition problems because of the keep alive messages used by the Apps. These small data transfers dont let the UE go in the IDLE state. If they do then whole raft of signalling has to occur again for the UE to go to CELL_FACH or CELL_DCH. In another post Martin also pointed out the sluggishness caused by the UE in CELL_FACH state.


Mike Thelander of the Signals Research Group presented similar story in the recently concluded LTE World Summit. It can be seen from the figure above that moving from IDLE to CELL_DCH is 1-3secs whereas FACH to DCH is 500ms.

In case if some Apps are running in the background, they can be using these keep alive messages or background messages which may be very useful on the PC but for the Mobiles, these could cause unnecessary state transitions which means lots of signalling overhead.

The Apps creators have realised this problem and are working with the Phone manufacturers to optimise their messaging. For example in case of some Apps on mobiles the keep alive message has been changed from 20 seconds to 5 mins.

3GPP also realised this problem quite a while back and for this reason in Release-7 two new features were added in HSPA+. One was Continuous Packet Connectivity (CPC) and the other was Enhanced CELL_FACH. In Release-8 for HSPA+, these features were added in UL direction as well. The sole aim of these features were to reduce the time it would take to transit to CELL_DCH. Since CPC increases the cell capacity as well, more users can now be put in CELL_FACH instead of being sent to IDLE.

An interesting thing in case of LTE is that the RRC states have been simplified to just two states as shown here. The states are IDLE and CONNECTED. The intention for LTE is that all the users can be left in the CONNECTED state and so unnecessary signalling and time spent on transitioning can be reduced.

The preliminary results from the trials (as can also be seen from here) that were discussed in the LTE World Summit clearly show that LTE leads to a capacity increase by 4 times (in the same BW) and also allow very low latency. I am sure that enough tests with real life applications like Skype, Fring and Yahoo IM have not been done but I am hopeful of the positive outcome.

HSDPA Code Tree

How often does it happen that people ask you questions you know the answer to but cant recall the complete details. A similar thing happened when a colleague asked me about why only 15 codes why HS-PDSCH and what happens to the 16th code.

Here is a picture which is from Qualcomm Whitepaper (available here) which is self explanatory.

Inter-Layer Communication Primitives

IEEE defines service primitives that are used for communication between different layers in a protocol stack. There are 4 types of service primitives as can be seen in the diagram above and are described below:

Request: This is sent by the initiating side and from a higher layer to a lower layer. For example when RRC wants to send a message to peer RRC entity, it sends an RLC Data Request to RLC.

Indication: This primitive on the receiving entity is passed from Layer N to the layer above (N+1). For example when RLC entity receives MAC data from MAC and its addressed to RRC, it sends RLC Data Ind to the RRC.

Response: This is the response to the Indication on the receiving entity. So in our example case, RLC Data Resp would be sent by RRC when it receives RLC Data Ind.

Confirm: This is used as a reply in the sending entity as the lower layer conveys the result of one or more previous request primitives. The confirm will generally contain status code indicating success or failure of the procedure. In our example, RLC Data Cnf will be sent by RLC as a response to RLC Data Req.

Flexible RLC in Release 7 and Release 8

In R99, RLC packets had to be relatively small to avoid the retransmission of very large packets in case of transmission errors. Another reason for the relatively small RLC packet size was the need to provide sufficiently small step sizes for adjusting the data rates for Release 99 channels.

The RLC packet size in Release 99 is not only small, but it is also fixed for Acknowledged Mode Data and there are just a limited number of block sizes in UM Data. This limitation is due to transport channel data rate limitations in Release 99. The RLC payload size is fixed to 40 bytes in Release 99 for Acknowledged Mode Data. The same RLC solution is applied to HSDPA Release 5 and HSUPA Release 6 as well: the 40-byte packets are transmitted from RNC to the base station for HSDPA. An additional confi guration option to use an 80-byte RLC packet size was introduced in Release 5 to avoid extensive RLC protocol overhead, L2 processing and RLC transmission window stalling. With the 2 ms TTI used with HSDPA this leads to possible data rates being multiples of 160 kbps and 320 kbps respectively.

As the data rates are further increased in Release 7, increasing the RLC packet size even further would significantly impact on the granularity of the data rates available for HSDPA scheduling and the possible minimum data rates.

3GPP HSDPA and HSUPA allow the optimization of the L2 operation since L1 retransmissions are used and the probability of L2 retransmissions is very low. Also, the Release 99 transport channel limitation does not apply to HSDPA/HSUPA since the L2 block sizes are independent of the transport formats. Therefore, it is possible to use fl exible and considerably larger RLC sizes and introduce segmentation to the Medium Access Control (MAC) layer in the base station.

This optimization is included for downlink in Release 7 and for uplink in Release 8 and it is called flexible RLC and MAC segmentation solution. The RLC block size in fl exible RLC solution can be as large as an Internet Protocol (IP) packet, which is typically 1500 bytes for download. There is no need for packet segmentation in RNC. By introducing the segmentation to the MAC, the MAC can perform the segmentation of the large RLC PDU based on physical layer requirements when needed. The fl exible RLC concept in downlink is illustrated in Figure above.


There is a lot of interesting information in R&S presentation on HSPA. See here.

Main source of the content above and for further information see: LTE for UMTS: OFDMA and SC-FDMA Based Radio Access

Enhanced UL for CELL_FACH state in Release 8

Users should always be kept in the state that gives the best trade-off between data rate availability, latency, battery consumption and usage of network resources. As a complement to the data rate enhancements made to the dedicated state (CELL_DCH), 3GPP has also made significant enhancements to the common states (URA_PCH, CELL_PCH and CELL_FACH). Release 7 introduced HSDPA mechanisms in the common states in order to improve their data rates, latency and code usage. Release 8 introduces corresponding enhancements in the uplink, allowing base stations to configure and dynamically manage up to 32 common Enhanced Uplink resources in each cell.



This enhancement improves latency and data rates for keep-alive messages (for example, from VPN or messenger applications) as well as web-browsing events, providing a seamless transition from EUL in common state to EUL in dedicated state.

As a further improvement of the CELL_FACH state, Release 8 introduces discontinuous reception (DRX), which significantly reduces battery consumption. DRX is now supported in all common and dedicated states.



Enhanced FACH and RACH bring a few performance benefits:

• RACH and FACH data rates can be increased beyond 1 Mbps. The end user could get immediate access to relatively high data rates without the latency of channel allocation.
• The state transition from Cell_FACH to Cell_DCH would be practically seamless. Once the network resources for the channel allocation are available, a seamless transition can take place to Cell_DCH since the physical channel is not changed.
• Unnecessary state transitions to Cell_DCH can be avoided when more data can be transmitted in Cell_FACH state. Many applications create some background traffic that is today carried on Cell_DCH. Therefore, Enhanced RACH and FACH can reduce the channel element consumption in NodeB.
• Discontinuous reception could be used in Cell_FACH to reduce the power consumption. The discontinuous reception can be implemented since Enhanced FACH uses short 2 ms TTI instead of 10 ms as in Release 99. The discontinuous reception in Cell_FACH state is introduced in 3GPP Release 8.

For more information see: LTE for UMTS: OFDMA and SC-FDMA Based Radio Access

HSPA Functions and Benefits

Very interesting diagram summarising HSPA Functions and Benefits

Source: 3G Americas Whitepaper, HSPA to LTE-Advanced: 3GPP Broadband Evolution to IMT-Advanced (4G)

Orthogonality and non orthogonality

Multiple access (MA) is a basic function in wireless cellular systems. Generally speaking, MA techniques can be classified into orthogonal and non-orthogonal approaches. In orthogonal approaches, signals from different users are orthogonal to each other, i.e., their cross correlation is zero, which can be achieved by time division multiple-access (TDMA), frequency-division multiple-access (FDMA) and orthogonal-frequency division multiple-access (OFDMA). Non-orthogonal schemes allow non-zero cross correlation among the signals from different users, such as in random waveform code-division multiple-access (CDMA), trellis-coded multiple-access (TCMA) and interleave-division multiple-access (IDMA).

First and second generation cellular systems are dominated by orthogonal MA approaches. The main advantage of these approaches is the avoidance of intra-cell interference. However, careful cell planning is necessary in these systems to curtail cross-cell interference. In particular, sufficient distance must exist between re-used channels, resulting in reduced cellular spectral efficiency.

Non-orthogonal CDMA techniques have been adopted in second and third generation cellular systems (e.g. CDMA2000 and uplink WCDMA). Compared with its orthogonal counterparts, CDMA is more robust against fading and cross-cell interference, but is prone to intracell interference. Due to its spread-spectrum nature, CDMA is inconvenient for data services (e.g., wireless local area networks (WLANs) and 3GPP high speed uplink/downlink packet access (HSUPA/HSDPA) standard) that require high single-user rates.

Communication services can be classified into delay sensitive and insensitive ones. A typical example of a delay-insensitive service is email. Typical examples of delay-sensitive services include speech and video applications. For delay insensitive services, rate constraints are relatively relaxed for individual users and maximizing the throughput by orthogonal methods is a common strategy. The maximum throughput can be achieved by a one-user transmission policy, where only the user with the largest channel gain is allowed to transmit. This implies time domain orthogonality as adopted in many WLANs. For delay-sensitive services, on the other hand, each user must transmit a certain amount of information within a certain period and maximizing the throughput is no longer an appropriate strategy. Rate constraints must be considered in this case.

CDMA is the most well known non-orthogonal technique. The main advantages of CDMA are its robustness against fading and cross-cell interference, and its flexibility in asynchronous transmission environments.

An uplink data transfer mechanism in the HSUPA is provided by physical HSUPA channels, such as an Enhanced Dedicated Physical Data Channel (E-DPDCH), implemented on top of Wideband Code Division Multiple Access (WCDMA) uplink physical data channels such as a Dedicated Physical Control Channel (DPCCH) and a Dedicated Physical Data Channel (DPDCH), thus sharing radio resources, such as power resources, with the WCDMA uplink physical data channels. The sharing of the radio resources results in inflexibility in radio resource allocation to the physical HSUPA channels and the WCDMA physical data channels. In CDMA, which is a non-orthogonal multiple access scheme, the signals from different users within the same cell interfere with one another. This type of interference is known as the intra-cell interference. In addition, the base station also receives the interference from the users transmitting in neighbouring cells. This is known as the inter-cell interference.

Uplink power control is typically intended to control the received signal power from the active user equipments (UEs) to the base as well as the rise-over-thermal (RoT), which is a measure of the total interference (intra- and inter-cell) relative to the thermal noise. In systems such as HSUPA, fast power control is required due to the fast fluctuation in multi-user (intra-cell) interference. This fast fluctuation will otherwise result in the well-known near-far problem. Moreover, as uplink transmission in an HSUPA system is not orthogonal, the signal from each transmitting UE is subject to interference from another transmitting UE. If the signal strength of UEs varies substantially, a stronger UE (for example, a UE in favourable channel conditions experiencing a power boost due to constructive short term channel fading such as Rayleigh fading) may completely overwhelm the signal of a weaker UE (with signal experiencing attenuation due to short term fading). To mitigate this problem, fast power control has been considered previously in the art where fast power control commands are transmitted from a base station to each UE to set the power of uplink transmission.

When an orthogonal multiple access scheme such as Single-Carrier Frequency Division Multiple Access (SC-FDMA), which includes interleaved and localized Frequency Division Multiple Access (FDMA) or Orthogonal Frequency Division Multiple Access (OFDMA), is used, multi-user interference is not present for low mobility and small for moderate mobility. This is the case for the next generation UMTS i.e. LTE system. LTE system employs SC-FDMA in uplink and OFDMA in downlink. As a result in the case of LTE, the fluctuation in the total interference only comes from inter-cell interference and thermal noise which tends to be slower. While fast power control can be utilized, it can be argued that its advantage is minimal. Hence, only slow power control is needed for orthogonal multiple access schemes.

Fundamental difference between HSDPA and HSUPA

It has been long time since HSDPA and HSUPA came into existence. Untill now we have read and implemented many features related to HSDPA and HSUPA. However following are the basic difference between HSDPA and HSUPA:

• In the downlink, the shared resource is transmission power and the code space, both of which are located in one central node, the NodeB. In the uplink, the shared resource is the amount of allowed uplink interference, which depends on the transmission power of multiple distributed nodes, the UEs.
• The scheduler and the transmission buffers are located in the same node in the downlink, while in the uplink the scheduler is located in the NodeB while the data buffers are distributed in the UEs. Hence, the UEs need to signal buffer status information to the scheduler.
• The WCDMA uplink, also with Enhanced Uplink, is inherently non-orthogonal, and subject to interference between uplink transmissions within the same cell. This is in contrast to the downlink, where different transmitted channels are orthogonal. Fast power control is therefore essential for the uplink to handle the near-far problem. The E-DCH is transmitted with a power offset relative to the power-controlled uplink control channel and by adjusting the maximum allowed power offset, the scheduler can control the E-DCH data rate. This is in contrast to HSDPA, where a (more or less) constant transmission power with rate adaptation is used.
• Soft handover is supported by the E-DCH. Receiving data from a terminal in multiple cells is fundamentally beneficial as it provides diversity, while transmission from multiple cells in case of HSDPA is cumbersome and with questionable benefits as discussed in the previous chapter. Soft handover also implies power control by multiple cells, which is necessary to limit the amount of interference generated in neighbouring cells and to maintain backward compatibility and coexistence with UE not using the E-DCH for data transmission.
• In the downlink, higher-order modulation, which trades power efficiency for bandwidth efficiency, is useful to provide high data rates in some situations, for example when the scheduler has assigned a small number of channelization codes for a transmission but the amount of available transmission power is relatively high. The situation in the uplink is different; there is no need to share channelization codes between users and the channel coding rates are therefore typically lower than for the downlink. Hence, unlike the downlink, higher order modulation is less useful in the uplink macro-cells and therefore not part of the first release of enhanced uplink.

Implementation of CQI Reporting in HSPA

In HSDPA the channel quality indicator is a measure of the mobile channel which is send regularly from the UE to the Node B. These measurements are used to adapt modulation and coding for the corresponding UE and it can be also used for the scheduling algorithms.

The CQI measurement is implemented in the HSPA module and the measurement interval as well as the influence of measurement errors can be parameterised. The results can be given in form of maps or in a statistical manner as histogram for each cell.

Information about the instantaneous channel quality at the UE is typically obtained through a 5-bit Channel-Quality Indicator (CQI) in HS-SCCH, which each UE feed back to the NodeB at regular intervals. The CQI is calculated at the UE based on the signal-to-noise ratio of the received common pilot. Instead of expressing the CQI as a received signal quality, the CQI is expressed as a recommended transport-block size, taking into account also the receiver performance.

The reason for not reporting an explicit channel-quality measure is that different UEs might support different data rates in identical environments, depending on the exact receiver implementation. By reporting the data rate rather than an explicit channel-quality measure, the fact that a UE has a relatively better receiver can be utilized to provide better service (higher data rates) to such a UE. It is interesting to note that this provides a benefit with advanced receiver structures for the end user.

This is appropriate as the quantity of relevance is the instantaneous data rate a terminal can support rather than the channel quality alone. Hence, a terminal with a more advanced receiver, being able to receive data at a higher rate at the same channel quality, will report a larger CQI than a terminal with a less advanced receiver, all other conditions being identical.

Each 5-bit CQI value corresponds to a given transport-block size, modulation scheme, and number of channelization codes. Different tables are used for different UE categories as a UE shall not report a CQI exceeding its capabilities. For example, a UE only supporting 5 codes shall not report a CQI corresponding to 15 codes, while a 15-code UE may do so. Therefore, power

offsets are used for channel qualities exceeding the UE capabilities. A power offset of x dB indicates that the UE can receive a certain transport-block size, but at x dB lower transmission power than the CQI report was based upon. UEs belonging to category 1–6 can only receive up to 5 HS-DSCH channelization codes and therefore must use a power offset for the highest CQI values, while category 10 UEs are able to receive up to 15 codes.

The CQI values listed are sorted in ascending order and the UE shall report the highest CQI for which transmission with parameters corresponding to the CQI result in a block error probability not exceeding 10%.

Specifying which interval the CQI relates to allows the NodeB to track changes in the channel quality between the CQI reports by using the power control commands for the associated downlink (F-) DPCH. The rate of the channel-quality reporting is configurable in the range of one report per 2–160 ms. The CQI reporting can also be switched off completely.

In addition to the instantaneous channel quality, the scheduler implementation in the NodeB should typically also take buffer status and priority levels into account before finalising the data rate for the UE. Obviously UEs for which there is no data awaiting transmission should not be scheduled. There could also be data that is important to transmit within a certain maximum delay, regardless of the channel conditions. One important example hereof is RRC signalling, for example, related to cell change in order to support mobility, which should be delivered to the UE as soon as possible. Another example, although not as time critical as RRC signalling, is streaming services, which has an upper limit on the acceptable delay of a packet to ensure a constant average data rate. To support priority handling in the scheduling decision, a set of priority queues is defined into which the data is inserted according to the priority of the data. The scheduler selects data from these priority queues for transmission based on the channel conditions, the priority of the queue, and any other relevant information.