Some clarifications on the misconceptions of EcNo and why it is not used in HSDPA Drive Test to evaluate link quality.

1. Why CQI?Some Questions answered

2. Why CQI Several times we’ve asked ourselves these questions: Q: Why do we get poor Ec/No when conducting HSDPA Drive Test? Q: What is the calculations behind Ec/No going low when HSDPA session is initiated? Q: Why do we consider CQI as a Quality Metric when conducting HSDPA Drive Test instead of Ec/No? Take a look at the following slides

3. Why CQI Note the CQI value is 27 (good) while EcNo is about to explode (-12db)

4. Lets start by establishing some basics: Ec/No for most of you is a Quality & Capacity measuring Metric. It generally gives us the idea how good or bad the link Quality is. However, by definition, its quite confusing – CPICH Ec/No = Pilot channel quality energy per chip over total received power spectral density OR RSCP = RSSI + Ec/No EcNo = RSCP - RSSI Let’s go back to developing Basics WHAT?!?!? Ok…..

5. RSCP: The Received Signal Code Power: That's the power level the pilot channel of a cell is received with and usually expressed in dBm (mW on a logarithmic scale). With this parameter, different cells using the same carrier can be compared and handover or cell reselection decisions can be taken. RSSI: In UMTS that's the signal power over the complete 5 MHz carrier which includes all components received, including the signals from the current and neighboring cells on the same frequency. In an isolated cell having only CPICH power to transmit and with none of the other channels utilized, the effective RSSI as seen by the UE would be (fraction of) PCPICH power. However, the RSSI changes when the 5Mhz carrier will carry power for other dedicated or control channels. In that case, the effective power perceived by the UE would be the transmitted power of the serving cell + the power transmitted by neighbor cells over the same carrier frequency. EcNo: That's the received energy per chip (Ec) of the pilot channel divided by the total noise power density (No). In other words the EcNo is the RSCP divided by the RSSI. And again in other words: The better this value the better can a signal of a cell be distinguished from the overall noise. The EcNo is usually expressed in dB as it's a relative value. The value is negative as the RSCP is smaller than the total received power. As the RSCP this value can be used to compare different cells on the same carrier and handover or cell reselection decisions can be taken. Let’s go back to developing Basics

6. It is believed that Ec/No is an expression of signal-to-noise ratio and is therefore, an indicator of signal quality. IT IS NOT (just that)! The “N0” consists of three things: interfering (non-orthogonal) power, thermal noise, and, non-interfering (orthogonal) power. In situations where N0 consists mainly of "in-cell" power (e.g., during HS-DSCH bursts), Ec/N0 can be "low", whereas the objective signal quality is high. This will be shown in the following slides Misconceptions about Ec/No

7. Further, Ec is an expression of power in the CPICH, not in the downlink DPxCH. Though the dynamic range of downlink DPxCH power is often expressed relative to the CPICH, downlink power control will do its best to ensure that downlink SNR is whatever it needs to be to meet the transport channel quality target. The point being: The traffic channel SNR has almost nothing to do with pilot Ec/N0. It is also to be noted that UE is only capable of receiving Pilot RSCP and RSSI on the carrier. EcNo is calculated by the UE by the formula (RSCP-RSSI) and sent in the UL. Misconceptions about Ec/No

8. Don’t be shy. I know we’re still Confused …..  Basics…….

9. Then, Ec/No for a UE is The measure of PCPICH (code power) over Total Wideband Power on that particular carrier Measure of PCPICH = RSCP dBm (right?) Measure of Total Wideband power = RSSI dBm So our Ec/No will become Ec/No = RSCP / RSSI (cannot divide dBm values so we have to apply logarithmic rules) Ec/No = RSCP – RSSI (db) (dBm – dBm = value in dB) For the sake of simplicity, just imagine that you’re in a test-bed, being served by a single cell

10. Hope you’re able to make a picture of how Ec/No works……. If yes, proceed further to find what happens when an HSDPA session is initiated If not, I suggest you discuss this concept first before proceeding. Are you there yet…?

11. Let’s take a look at how the POWER of a CELL is distributed among different channels For simplicity, check the diagram below. Notice that HSDPA uses the REMAINING power of the cell after power is allocated to Common and Dedicated channels (check more detailed figure in the next slide) HSDPA Cell Power Distribution Dedicated channels (power controlled) Total available cell power Common channels (not power controlled)

12. Cell Power Distribution The figures highlight the following points: Common Channels have the highest priority for Power. Their power is fixed DCH come second in priority – Power control is applicable All that remains after DCH utilization can be used up for HSDPA power So if a cell is less loaded, a single UE HS session can utilize the whole HSDPA power space which is lying unused.

13. Now imagine yourself conducting a test in Test Bed on a cell having ZERO DCH or HS Traffic. Assuming only 10% cell power is dedicated to CPICH (33dBm = 2watt) Cell MAXTXPOWER is 20W or 43dBm Ideally, if you have NO traffic on this cell, your UE will calculate the CPICH Ec/No in the following manner Ec/No = 10 log (CPICH power / Total Transmit power) Ec/No = 10 log (2w/2w) = 10 log (1) = 0 dB (VERY high, but that’s IDEAL!!!) OR (in dBm) Ec/No = RCSP – RSSI Ec/No = 33 (dBm) – 33 (dBm) = 0 * Ec/No Computation – No Load

14. Ec/No Computation – One HSDPA user Now assume you start an HS session.Since there is no DCH traffic, all the REMNANT power is given to HSDPA Cell Transmit power is now 2W for CPICH and REMNANT power (18W) for HSDPA – so ideally, Cell is utilizing almost 100% power So UE will measure the Ec/No in the following manner Ec/No = 10 log (CPICH power / Total Transmit Power) Ec/No = 10 log (2W / 20W) Ec/No = 10 log (0.1) = -10dB (Very low – again very IDEAL!) OR (in dBm) Ec/No = RCSP – RSSI Ec/No = 33 (dBm) – 43 (dBm) = -10 dB

15. Why CQI From above computations, we establish the following: In Idle mode and with no resources allocated, a UE will measure as low as 0 dB Ec/No In HS Mode and with no resources allocated on DCH, a UE will measure as low as -10dB. It cannot report ANY better than -10dB Hence, that means, in a live network where resources of cell are shared between many users, and Interference from other cells also plays its part, the Ec/No will always give a FALSE value for an HSDPA user. And it will show a very poor value What you should be asking yourself: What happens when more users join in to share the REMNANT power? Will Ec/No improve or degrade further with addition of DCH or HS user?

16. So, we establish that, The CPICH Ec/No measurements collected during HSDPA Transmission could mask the true radio conditions To avoid this, we have to monitor the Ec/No in IDLE mode only, which is a true reflection of Interference and coverage problems HSDPA session adds to the overall load of the cell which is taken into consideration during computing Ec/No. As calculations above proved that this does not, necessarily, depict the TRUE radio Conditions. So, what is the ALTERNATIVE to Ec/No in HS session? Its called Channel Quality Indication CQI Why CQI

17. What is CQI HSDPA utilizes link adaptation techniques to substitute power-control and variable spreading factor The HS-DSCH link-adaptation algorithm at the Node-B is very dynamic, and adjusts the transmit bit rate on the HS-DSCH every 2-ms TTI. The UE periodically sends a CQI to the serving HS-DSCH cell on the uplink high-speed dedicated physical control channel (HS-DPCCH) The CQI tells the NodeB scheduler, the data rate the UE expects to be able to receive at a given point in time. NOTE: The HSDPA system defines a different CQI mapping table for different categories of UEs. The category is determined according to the capability of UE In addition to CQI, UE also sends HARQ on HS-DPCCH

18. Below figure shows the steps how CQI helps in determining the initial throughput that a UE terminal demands from the system STEPS SINRis used to evaluate the channel quality as observed by the receiver, where a standard single antenna Rake is used because it is the most common in the SISO HSDPA terminals. CQI value for a given SINR is done via a linear mapping, as shown in figure Contd….. HSDPA – CQI involvement

19. HSDPA – CQI involvement STEPS The CQI values are used by the link adaptation algorithm at the Node-B. Every CQI value reported corresponds to the Transport Block Size (TBS) that can be granted on a particular Modulation type and Number of codes Example of a UE category 10 CQI values from 0 to 30 are shown in the figure. Other UE categories are shown in the following slide The CQI is further defined as the TBS that can be supported with a BLER no greater than 10% -

20. STEPS While CQI is sending UE’s requirement to the system in UL, there is BLER calculation going on and UL HARQ (Hybrid Auto Repeat Request) mechanism is helping in maintaining the BLER to below 10%. So the system (NodeB) is essentially calculating the Data Rate to be scheduled to the user based on CQI reports and BLER which it receives from the UE So DATA RATE (bps) = TBS (bits)/ TTI (sec) * (1-BLER) HSDPA – CQI involvement

21. HSDPA – UE Categories 1-12

22. Thank you for reading