For a wireless communication device to work properly, it should meet following hardware requirement
i) The device should transmit the signal which is strong enough power to make it sure it reaches the other party of the communication.
ii) The device should not transmit the signal which is so strong that it interfere the communication between other parties.
iii) The device should transmit the signal with good enough quality which can be decoded/corrected by the other party.
iv) The device should transmit the signal in the exact frequency that has been allocated for the communication.
v) The device should not generate any noise out side of the frequency area that has been allocated for the device.
If any of these condition deviate too much from the specification, the device cannot communicate with the other party or let some other device to communicate. In terms of measurement equipment, item i) and ii) belong to "power measurement", item iii) is related to "Modulation Analysis" and item iv) falls into "Frequency Error measurement". Item v) is also a kind of "power measurement", but the measurement area in frequency domain is different from item i) & item ii). Anyway if you have any equipment that can perform the following three measurement for your communication technology, you can do the most critical part of transmission path.
a) Power Measurement
b) Modulation Analysis
c) Frequency Error Measurement
Now let's think about the recieve path measurement. What would be the most important reciever characteristics for the communication device ?
i) The reciever must be able to decode successfuly the signal coming from a transmitter even though the signal strength is very low.
ii) The reciever must be able to decode successfuly the signal coming from a transmitter even when there are a certain level of noise around the signal.
In terms of measurement logic, item i) and ii) are the same. Equipment sends a pattern of the known signal and let the reciever decode it and compare the original signal from the equipment and the decoded signal by reciever and how much different they are. The more different they are, the poorer reciever quality it is. We call this method "BER(Bit Error Rate) measurement". Item i) measures BER when the input signal to the device is very low and Item ii) measures BER when there are noise to the input signal.
Before we go forward to LTE measurement, pick any technology you are already familiar with and make a list of measurement on your test plan and try to map those items with the measurement principles I described above. Once you are familiar with this mapping, you will understand LTE measurement items more easily.
LTE RF Measurement Items
Now let's look a little bit detail into LTE RF measurement. First thing I have done is to make a list of measurement items from 3GPP 36.521-1 and try to map my measurement principles with each of the measurement items.
Here goes the Transmitter measurement items first. You see a lot of "Power Measurement" and some of "Modulation Analysis". Why do we have so many different power measurement and so many different Modulation Analysis. How do they differ from each other ? This is the question you have to find answers on your own. The answer itself is described in 3GPP 36.521-1 but the question is how much I can understand what is described there just by reading it.
The first step would be to read "Test Purpose", "Initial Condition", "Test Procedure" section of each test case as often as possible and try at least to be familiar to each test case.
Here goes the reciever measurement items.
Snapshots of LTE Uplink Signals for RF Testing
As I mentioned earlier, it is not easy to understand all the details of LTE RF Measurement just by reading the specification. I have read the test case purpose, "Initial Condition", "Test Procedure" over and over.. but still everything is vague. As I try to get more into details, the first obstables that blocks me is a lot of complicated tables describing the test condition. Of course we saw this kind of tables in other technology specification like CDMA, WCDMA but it seems the tables for LTE measurement looks bigger and more complicated. So I decided to see some of the signal patterns described in the specification on spectrum analyzer so that I can get some intuitive idea of the overall RF characteristics of each condition.
Even though we have new technology every couple of years and LTE is new to many people, RF test and measurement technology have a lot in common with other wireless communication technology. If you had experience with any wireless technology, eg CDMA, GSM, WCDMA, Bluetooth, WLAN, you may find the common logics in LTE.
One of the biggest challenges in LTE measurement for UE development or test engineer would be that there are too many sub tests with too many different parameter settings.Before I get into details, I want to briefly skim through overall RF measurement from C2K.
I don't have much experience with C2K measurement, but with only a little experience I could tell there are much fewer measurement items in this area comparing to WCDMA/HSDPA and even comparing to GSM/GPRS. As far as I remember, following is allmost all that I did for C2K.
i) Total Channel Power
ii) CDP (Code Domain Power)
iii) Rho
iv) Spectrum Emission
v) ACLR
vi) OBW (Occupied Bandwidth)
But the items listed above is more than what I experienced in C2K. For conformance, I think we may have to go through all of these items. But since C2K is very mature technology now, in the RF part developmental stage we wouldn't go through all of these items. In an extreme case that I heard of was "just measure total power, if there is no problem with it. usually no problem with other parts".
Now let's look into WCDMA. For WCDMA R99 (Non HSPA), If I briefly put the list,
i) Max Power
ii) Min Power
iii) On/Off Power
iv) RACH Power
v) EVM
vi) Spectrum Emission
vii) ACLR
viii) OBW (Occupied Bandwidth)
Just in terms of list, it doesn't look like much difference from C2K. But practically the engineer would meet various characteristics which may look quite different from C2K. The first thing we can think of is that the channel bandwith get tripled compared to C2K and this would introduce a lot of complication in RF design. Another issue is RACH process in WCDMA is more complicated than probing process in C2K and add a couple of important test steps.
Now let's look further into HSDPA. You may think HSDPA would not be much different from R99 in terms of Uplink measurement because HSDPA is only for downlink data rate. It is true in terms of high level protocol, but in physical/RF layer an important factor was added to uplink in HSDPA. It is HS-DPCCH. HS-DPCCH is for UE to report CQI and ACK/NACK to BTS. The problem is that even with this additional channels the UE has to maitain the total uplink power as before. So the UE recalculate/rearrange each of the physical channel power. So if you look at the RF conformance test case list, you would not find much difference in terms of test case items but you would find quite a many of sub items were added to the existing test case due to the introduction of HSDPCCH. (If you want to go into further detail, open up 3GPP 34.121 and find the test cases with the keyword "HSDPCCH" in the test title).
Going one step further into HSUPA, you also find no such a big difference in terms of measurement items. But as in HSDPA case, a new physical channel was introduced and it is called E-DPCH. Even with this additional channel, UE also have to maintain the total channel power as in R99. So, as you may guess, UE has to recalculate/rearrange each of physical channel powers. As a result, we would get a couple of additional sub-items added to RF testing.
Finally.. let's think about LTE. What is the biggest difference between LTE and C2K/WCDMA/HSPA in terms of PHY/RF layer ? It would be OFDM. Yes, it is. What kind of additional measurement items would be introduced to RF testing due to the OFDM ? Since OFDM is made up of a lot of sub carrier with very narrow bandwidth, we have to measure most of the characteristics listed above for each OFDM subcarrier. But if we do all of the items for each of the sub carriers, it would take one full day just for one item. Another big difference would be that LTE specification allow many different type of system bandwidth whereas in C2K/WCDMA, the system bandwidth is always same. It means you have to measure the whole set of test items for multiple different system bandwidth which multiplies the measurement time and parameter settings in measurement equipment.Based on the LTE specification, an LTE system bandwidth can be any of 1.4 Mhz, 3 Mhz, 5 Mhz, 10 Mhz, 15 Mhz, 20 Mhz whereas C2K can only have single bandwidth of 1.28 and WCDMA can only have single bandwidth of 3.84. Of course, a specific system operator would use only one of the bandwidth in their network but Mobile device manufacturer should design the UE which support all of these bandwidth.On top of this, there is another factors to make LTE test even more complex especially for mobile phone design/test. It is the fact that a real bandwidth being used at a specific time can change dinamically.
One intuitive example is shown in the following measurement screen. This the RF signal captured for LTE call connection and data transfer. When you initiate a call, the mobile device would go through the protocol sequence for call setup and then data fraffic would start. If you see at the bottom of the screen (spectrogram) of the measurement screen, you would notice that frequency allocation (bandwidth being used) during this period changes. In this screen, the frequence allocation for data traffic does not change, but in live network this bandwidth would change dynamically.
What is the implication of these multiple system bandwith and dynamic bandwidth change to Mobile phone designer and the test engineer ? For designers, the biggest issues would be how to optimize various kinds of design parameters to be best fit for all of these bands. For test engineers, the biggest issue would be huge number of the test cases they have to go through.
Final outcome of all these considerations on multiple bandwidth and dynamic bandwidth change can be examplified as a table shown below. This is a table for only one test case. See all those different system bandwith you have to cover. Different RB allocations is for dynamic frequency allocation that I mentioned above. In LTE, for every test case you would have this kind of tables and this will be huge headache to designers and test engineers. 
