The HS-DSCH downlink channel is shared between users using channel-dependent scheduling to make the best use of available radio conditions. Each user device periodically transmits an indication of the downlink signal quality, as often as 500 times per second. Using this information from all devices, the base station decides which users will be sent data on the next 2 ms frame and how much data should be sent for each user. More data can be sent to users which report high downlink signal quality.

The amount of the channelisation code tree, and thus network bandwidth, allocated to HSDPA users is determined by the network. The allocation is "semi-static" in that it can be modified while the network is operating, but not on a frame-by-frame basis. This allocation represents a trade-off between bandwidth allocated for HSDPA users, versus that for voice and non-HSDPA data users. When the base station decides which users will receive data on the next frame, it also decides which channelisation codes will be used for each user. This information is sent to the user devices over one or more HSDPA "scheduling channels"; these channels are not part of the HSDPA allocation previously mentioned, but are allocated separately. Thus, for a given 2 ms frame, data may be sent to a number of users simultaneously, using different channelisation codes. The maximum number of users to receive data on a given 2 ms frame is determined by the number of allocated channelisation codes. By contrast, in CDMA2000 1xEV-DO, data is sent to only one user at a time.

36.213 series (2) - Random Access

RACH is the interface between non-synchronized UEs and eNB.

Difference between LTE and WCDMA on RACH:

- RACH can't carry any user data, just pre-ambles. One exception is the scheduling request (SR), if the UE doesn't have any uplink resource block to send SR, it is allowed to send SR via RACH

- RACH is orthogonal with other resources such as PUSCH/PUCCH

When to RACH?

- A UE in RRC_CONNECTED state, but not uplink synchronized, need to send uplink data or control message (e.g. meas. report, ACK/NACK for downlink data)

- A UE in RRC_CONNECTED statebeing handover from one cell to another

- A transition from RRC_IDLE to RRC_CONNECTED (e.g. initial access or TAU)

- Recovering from radio link failure

Contention based Random Access

step 1) UE randomly choose RACH preamble signature. One UE says : "hello, I am Zhang-san", but the other can also say the same. A collision could happen and a collision resolution is required. The random access message is sent according to PRACH configuration (acquired by UE via system information or handover command). Initial transmission power is based on open loop estimation with full path-loss compensation.

step 2) eNodeB sends RAR (random access response) on PDSCH, and addressed with an ID, the Random Access Radio Network Temporary Identifier (RA-RNTI). RA-RNTI identifies the Time-frequency slot in which the preamble was detected. If multiple UEs had collided by selecting the same signature at the same time, they would each receive the RAR.

The RAR gives UE a C-RNTI (temporately), timing information and an initial uplink resource grant.

step 3) UE(s) sends the uplink message (For e.g. a ACK/NACK, measurement report, RRC connect request, etc) on the uplink resource granted by RAR. If a collision happened during step 1), they are going to collide again at this step. Each UE shall sent its own UE identity or C-RNTI (if its has one) in this uplink message, so that proper contention resolution can be done in the next step. Note in this step, the C-RNTI or UE identity is now unique for each UE.

Step 4) Possibly, the uplink message from none of the UE can be decoded, then all UE will have to re-RACH (triggered by guard timer expiry). Or, if one UE is succesfully decoded, it will receive a contention resolution message address to it (by C-RNTI or UE identity, which echos the message from UE in step 3). Other UE will also hear this message but the message is not addressing to them! Other UEs will now realize there was a collision and quit current RACH procedure to start a new one.

Contention free Random Access

UE use pre-allocated preamble signature. eNB can reserve a set of preambles signatures to be used for contention free Random Access (total number of available preamble signatures in LTE is 64).

RSCP and Eb/No in WCDMA

RSCP is the CPICH chip power at receiver - it tells the path loss
EC/No is the Chip power to Noise ratio at receiver - it tells not only path loss, but also interference

On WCDMA - uplink and downlink has some essential difference:

Uplink:

For the W-CDMA up link, the interference is received at the Node B, and is thus the same for all links in that cell. Ideally then, the same received power at the Node B would be required for all links with the same SIR target. The received interference level will vary between the cells, but since the number of UEs can be expected to be large compared to the number of cells, and due to the similar placements of Node B antennas, this variation is relatively small. At low to medium loads, the noise rise at the Node B is typically between 0-5 dB, to be compared with the at least 50-60 dB range of the path loss. This means that the UE output power mainly will be determined by the path loss to the Node B.

Since uplink coverage is limited by the UE output power, the uplink quality will be strongly dependent on the path loss and hence on the CPICH RSCP.

downlink:

The W-CDMA downlink differs from the uplink in many ways.

• In the down link, the interference is received at the terminal, and is thus different for different locations within the network. The interference level is significantly higher at the cell border than at the cell centre, due to other cell interference. At the cell border, the other cell interference may be as much as three times larger than the interference from the home cell.
• Another aspect is that all the downlink power transmitted in the serving cell, except the own signal, will be recieved as interference. This interference will be subject to exactly the same coupling loss at every instant as the signal. The effect of this own cell interference will be reduced by the OVSF codes, which ideally cancel all own cell interference (except from P-SCH and S-SCH). Under realistic conditions the OVSF codes cancel about 50% of the interfering own cell power.
• Further, there is a floor of interference caused by the non-power controlled control channels, such as CPICH, CCPCH, P-SCH, S-SCH and so on. This means that even at low load there is a substantial amount of interference, many, many times larger than the receiver noise power.
• The total available transmission power is normally at least 100 times larger at the Node B than in the terminal.

As a consequence of these facts, the link power for the down link will not be proportional to the path loss between serving cell and terminal. Instead, the downlink power always has to match the interference level, which due to own cell interference is large even close to the cell centre.

Conclusion:

RSCP is suitable for Uplink quality evaluation, but EC/No is suitable for downlink quality.

TDMA, FDMA and CDMA

A novice to digital communication system may wonder what are the meanings of these terminologies.

Let's start from the "M" in the middle, Multiplexing. In telecom system, multiplexing is a term used to refer to a process where multiple signals are combined into one signal over a shared medium. The reason of multiplexing is to share a medium (or channel in terms of information theory), for example, a wire connection, or a radio frequency, which is often limited resource.

It should now easy to understand the initials: "T" - time, "F" - frequency or "C" - code are just different ways of multiplexing.

TDM - channel is divided into sub-channels by time slots. A typical example is the congestion reduction strategy in Beijing, today is my turn, tomorrow is yours :-)
FDM - channle is divided into sub-channels by frequencies - by modulating baseband signals onto different frequency carriers. A typical example is dividing the road into lanes, cars can drive along without interfering each other on diff. lanes (ideally :-)
CDM - channel is divided into sub-channels by scrambling codes. Difficult to understand? Not at all. An analogy to the problem of multiple access is a room (channel) in which people wish to communicate with each other. To avoid confusion, people could take turns speaking (time division), speak at different pitches (frequency division), or speak in different languages (code division). CDMA is analogous to the last example where people speaking the same language can understand each other, but not other people. Similarly, in radio CDMA, each group of users is given a shared code. Many codes occupy the same channel, but only users associated with a particular code can communicate.

What's the "A" then? - Relation to multiple access:

A multiplexing technique may be further extended into a multiple access method or channel access method, for example TDM into Time-division multiple access (TDMA) and statistical multiplexing into carrier sense multiple access (CSMA). A multiple access method makes it possible for several transmitters connected to the same physical medium to share its capacity. Multiplexing is provided by the physical layer, while multiple access also involves a media access control protocol, which is part of the data link layer.

36.213 series (1) - Power Control in LTE

Power control in LTE is not as critical as in WCDMA. This is due to two factors:

- In LTE uplink are orthogonal, therefore no intracell interference (at least in theory)

- LTE has the freedom to do frequency selective scheduling, which can be used to further reduce the interference from other cells.

However, power control still plays an important part in LTE because a proper power control can maximize system capacity, minimize inter-cell interference, also look after the fairness between cell centre and cell edge UEs.

UL Power Control

Different formulas are specified for UL PUSCH, PUCCH, SRS. Fundamentally the formulas can be considered as a sum of two main parts:

1. a basic open-loop operating point derived from static or semi-static parameters signaled by eNB
   
2. a dynamic offset updated from sub-frame to sub-frame.
   

Furthermore, the 1st part, basic open loop operating point is composed of two parts: A) a semi-static P0, which is a cell specific power level and a UE specific offset (to compensate for example UE's PL estimation error), and B) a open loop path-loss compensation.

Basic operating point = P0 + α ● PL

α is a factor to trade off the fairness of uplink scheduling against total cell capacity. Full path-loss compensation (alpha = 1) maximize fairness for cell edge UE, but causing more interference to neighboring cells. For PUCCH, full PL compensation is always used.

The 2^nd part, the dynamic offset is also composed of two parts, one part is MCS dependent power offset ∆_TF (TF is transport format), the other part the TPC command related power. The TPC can command UE to use a relative power offset comparing to its previous Tx power, or command UE to use a absolute power, regardless its previous Tx power.

Downlink Power Allocation

In the downlink, cell specific RS EPRE (Energy per RE) is semi-static, it only change when the eNB signals UE so. PDSCH EPRE is proportional to RS EPRE's, depends on PDSCH RE's position. If the PDSCH RE is on the same symbol where there is a RS (index 0, 4), ρ_B is used, otherwise ρ_A is used. The value of ρ_{A /} ρ_B is determined by P_A and P_B which are specified by eNB via MAC/RRC signaling, they are UE specific. For details, see 36.213 section 5.2. Below diagram is an example when P_A=0.8 (ρ_A = 0.8) and P_B=2 (ρ_{B /} ρ_A = 0.6)

MIMO in LTE (2) - Rx Diversity receiver combining techniques (1)

As mentioned in an earlier post about MIMO, in Rx diveristy, the receiver needs to combine multiple streams from different antenna into a single stream. The challenge here is how to use "effectively" the information from allmthe antennas. In fact, it is a just a matter of choose the appropriate weight for each received signals (see figure below). There are multiple ways:

Selective Combining (SC):
The receiver selects the antenna with the highest received signal power and ignore observations from the other antennas. w=[1,0,0,0...].

On flat-fading Rayleigh channel, if the channels from each receiving antenna is independent, then:
- with two receive antenna the effective big energy to noise ratio is 1.5 times Eb/N0
- with three receive antennas the effective bit energy to noise ratio is 1.833 times Eb/N0

- with four receive antennas, the effective bit energy to noise ratio is 2 times Eb/N0

MIMO in LTE (1)

MIMO is vital for LTE system's performance. From communication channel point of view, below transmission schemes can be used:

1) SISO - single input single output, in other word, non-MIMO

2) SIMO - A single data stream is input into the channel, multiple outputs from the channel, i.e. Rx Diversity. This is to mitigate the effect of multi-path fading by transmitting same information via multiple antennas. The multiple antennas created sufficiently de-correlated channels, therefore signals can be combined at the receiver end. MRC (Maximum Ratio Combining), SC(Selective Combining), EGC(Equal Gain Combining) or IRC(interference rejection combining) can be used at the receiver.

3) MISO - Multiple layer of data stream is input into the channel, single outputs from the channel, i.e. Tx diversity. Depends on the number of Tx antenna ports, SFBC(Space-Frequency Block Coded) or SFBC+FSTD(Frequency Switch Transmit Diversity) can be used for 2x or 4x Tx diversity respectively.

4) MIMO - Multiple-Input, Multiple-Output channel. At Tx side, serial to parallel mapping maps the data stream to multi-parallel streams, and at the Rx side, parallel to serial mapping combine the received data streams back into a single stream.