Because the available physical layer bandwidth is less than that of LTE (one carrier at 180 kHz), the physical layer procedure is also significantly different from before. Taking into account the NB-IoT's need for signal coverage enhancement, the 3GPP standard-setting team used "repeat transmission" to obtain the time domain The purpose of Coverage Enhancement (CE) is to achieve the following: In the standard specification, downlink transmission only allows cross-subframe scheduling, and uplink transmission supports inter-sub frame and cross Subcarrier schedule.
The NB-IoT adopts a centralized control mode to manage the radio resources required for data transmission between the eNB and the UE, and in the same way as the LTE system, the UE transmits or receives data according to the eNB directives respectively Downlink Assignment and Uplink Grant (Downlink Control Indicator, Downlink Control Indicator, DCI), the DCI format is used in the uplink part, the DCI N1 format is used in the downlink part, The Paging section uses the DCI N2 format.
The UE periodically monitors / monitors the area of the DCI transmission, that is, the narrowband Physical Downlink Control Channel (NPDCCH), also known as the Search Space, during its link with the base station. The UE receives its own DCI And then instructs to the corresponding data transmission area, that is, the narrowband physical downlink shared channel (NPDSCH) reception data according to the content thereof.
Based on the NB-IoT cross-subrack scheduling feature, LTE compares the position of the resource block and the number of resource blocks placed in the current Subframe data with the DCI to let the UE know that the data is located in the "frequency range NB-IoT notifies a scheduling delay parameter (called k0 standard) and resource block length, so that the UE knows that its own data is located in that "time interval." For NB-IoT Resource allocation and scheduling related parts, the following instructions.
Reduce the cost of DCI to make good use of search space to improve UE efficiency
As with LTE, a UE can obtain DCI information by searching for a specific interval, which can reduce irrelevant data in which the UE consumes unnecessary power consumption blindness.
In the NB-IoT, the search space is presented as a time interval. By pre-informing the UE related parameters such as the common search space (SIB2-NB) in the system information block type 2 (SIB2-NB) (Common Search Space) parameter and the RRC Connection Setup Message (UE Specific Search Space) parameter in a random access (Random Access) process, so that the UE can know in which time range there is a chance to blindly resolve itself DCI.
Under the standard specification, there is a great deal of flexibility in setting the search space, and the length of the search space can be selected according to the characteristics of the UE to be served. At the same time, in the same search space, the search space can be further selectively divided into 1, 2 , 4, 8 are used as the DCI transmission time of different UEs, and the divided length is the number of times that the DCI is repeatedly transmitted (T1CSS may select more division ratios).
As shown in FIG. 1, the search space in the blue area is Rmax (in this example, it is set as 8), and R = Rmax / 8, R = Rmax / 2, R = Rmax / 1, for example, respectively 1, 2, 4, 8, R is the number of repetitions, and the time block covered by R is called Candidate, The division you choose can also be considered as the number of candidate blocks in this search space.
Figure 1 search space alternative block diagram
In addition, the initial time position of different search spaces can be adjusted through parameter settings to avoid too many UEs being set in the same search space, resulting in limited UEs that can be served by the base station in unit time. Different parameters and ratio subscriptions will affect The number of UEs that can be served by the base station in a unit time and the effectiveness of the CEs can be adjusted and selected according to the scheduling policies currently decided on in practice.
After the UE camps on a certain base station, the UE monitors the corresponding search space according to the current online state. Currently, the standard defines Type1-NPDCCH common search space (T1CSS) and Type2-NPDCCH Search space (T2CSS), NPDCCH UE-specific search space (USS) three different uses of the search space:
T1CSS
When the UE is idle, the T1CSS is monitored based on the Default Paging Cycle (CP) agreed with the core network (CN). Since UEs at different CE levels are all provided with the same T1CSS length, the candidate blocks Dividing According to the standard, more choices can be made to meet the repeated transmission times of UEs at each CE level. When the UE searches for space in this paging cycle to solve the DCI and correctly receives the paging message, the UE performs a random access procedure , And the search space adjusted for T2CSS.
T2CSS
When the UE is not registered with the core network or is registered but in an idle state, if the UE intends to transmit data or receive a paging message from the base station, the UE starts to perform a random access procedure. Blind solution DCI is set according to T2CSS.
USS
When the UE finishes the random access procedure and enters the Connected state, the UE performs the search according to the USS parameter setting information obtained by the random access procedure until the state is switched to the idle state or the random access state again Corresponds to the search space switch.
Search / Transmit Job Diversified logical channels No clear rules for partitioning
Downstream channel
In NB-IoT system, exclude the necessary system / synchronization signals (such as NPBCH, NPSS, NSSS,
SIB-NB), there are two types of channels: NPDCCH and NPDSCH. However, there is no explicit time division rule between the two channels in terms of the entire NB-IoT system.
One of the reasons is that as mentioned above, the search space can be composed of very diverse combinations of different UEs and CEs regardless of the starting position or length; the second one is due to the scheduling delay mentioned in the next section, Therefore, in the downlink channel, we should look at the actual scheduling results to see the result of the division, which means that if the DCI is transmitted in a block time, the interval is used as the NPDCCH. If the downlink data is transmitted , This interval is used as NPDSCH.
Upstream channel
Compared with the downlink channel, the uplink channel division is simpler: the Preamble time block is sent as the NPRACH according to the random access operation set in the SIB2-NB, and the rest are all used as the NPUSCH.
The peculiar thing is, considering that the NB-IoT uplink supports cross-subcarrier scheduling, the scheduling shall further consider the resource allocation between subcarrier frequencies according to the selected NPUSCH format.
Delivering DCI / data schedule latency with limited bandwidth helps to balance efficiency
The 3GPP MAC protocol standard defines a Physical Downlink Control Channel Period (PP), that is, an interval from a start of a current search space to a start of a next search space, or a period of NB-IoT as an NPDCCH. 2, the blue area can be regarded as the NPDCCH of one / group UE and the white area is NPDSCH. The composition of the area is made up of standard set parameters and there are about 90 combinations. The selection of the period combination With scheduling strategy, CE consider the high flexibility to choose.
Figure 2 scheduling cycle and time diagram
One of the reasons that NB-IoT is scheduled through a cross-sub-frame is that the bandwidth defined by the system is small, and both the DCI and the data can not be transmitted at the same time. In a normal situation, for a transmission area Transport Block (TB) needs to be completed by multiple NPDSCHs, so how to deal with the time relationship between DCI and data is an NB-IoT-specific mechanism. Time series k0 plays the most important role.
When the UE obtains the k0 given by the base station after the DCI is solved from the candidate block, the UE waits for k0 time before starting the action of receiving the NPDSCH, while the k0 rule is in the uplink / downlink or in some specific messages For example, when the UE receives the DCI, it must wait for at least 4ms before receiving the NPDSCH and wait for at least 8 ms for the NPUSCH to be transmitted because the UE must have enough time to solve the DCI The message that comes with it, or the time it takes for UL / DL transfer and receive mode transitions.
Figure 3 is a schematic diagram of the maximum TB size and the MCS downlink interval set by the current standard Release 14 for uplink and downlink. From this figure, we can also calculate the maximum rate that the NB-IoT UE can reach in Release 14 Value.
Figure 3 NB-IoT scheduling diagram
The value of k0 of NB-IoT is chosen according to the fixed values specified in the standard file, so there is a lack of flexibility in the selection. In addition to the aforementioned diversity of search space starting positions and lengths, as well as the transfer The length of time spent on a TB will have an impact, so scheduling will be a challenging task to work on. The topics to be derived from it are subject to research and discussion. The following is an explanation of relevant scheduling issues.
The importance of augmenting UE service MAC scheduling with limited resources is increasing with time
Because NB-IoT supports multi-carrier transmission, different UEs can transmit on different carriers to expand the number of serving UEs. MAC scheduling and radio resource allocation play a crucial role.
An NB-IoT multi-carrier is divided into an anchor carrier and a non-anchor carrier. The anchor carrier is a carrier that the UE acquires system information and synchronization signals (NPSS / NSSS / NPBCH / SIB-NB) The anchor carrier can be regarded as a blank resource block if the system supports it. Since the anchor carrier serves as the relationship between the transmission system information and the synchronization signal, the information will be considered as the highest priority for resource occupation. Therefore, On NPDCCH and NPDSCH, if they encounter the above message transmission time, they need to delay the transmission.
In view of this, on the anchor carrier scheduling, we must consider the impact of these delays on resource scheduling; In addition, Release 14 standard specification compared to Release 13, you can random access procedures and paging procedures On non-anchor carriers, this approach increases the efficiency of the system but also relatively refers to the complexity of schedule allocation.
If the entire communication protocol and IoT service point of view, the transmission of an IoT message must be completed after the exchange of several messages on the NB-IoT system; if the message return initiated by the UE point of view, it must go through a complete randomization The procedure is to complete the transmission of an upper service data. The complete procedure initiated by this UE is called the so-called MO Procedure.
However, under many restrictions such as bandwidth limitation, scheduling period and search space, the base station has to decide on some topics under limited resources, such as uplink / downlink allocation ratio, UE resource allocation ratio, fairness, and the like Fig. 4 is a schematic diagram of a simplified time axis correspondence of MO Procedure for a single UE 10. In addition, considering the power saving mechanism such as the function of DRX status paging and the scheduling on multiple CE level settings, in the MAC resource allocation management It will be a big challenge.
Figure 4 NB-IoT schedule diagram
This article focuses on the NB-IoT MAC layer technology description, due to changes in the NB-IoT in the logic of resource allocation has obvious difference with its mother technology LTE, this section we focus on this part of the description and description.Although NB-IoT Is a relative simplification of LTE technology. However, in order to meet the simplification of time resources in exchange for the concept of frequency resources, the scheduling logic relatively more complex new issues need to be addressed; if the entire system of view, the row The approach will have greater impact on the overall performance. Considering the future standard set to add more scheduling issues, such as the 2-HARQ Process set in Release 14, MAC will play a key and important Character.