A new procedure developed by CEVA improves IoT devices' initial synchronisation process to reduce narrowband primary synchronisation signal detection latency.
Many IoT devices are designed to operate remotely and/or in inaccessible locations and for these and other reasons, low power consumption is a critical requirement. To reduce power consumption and improve battery life, devices are designed to operate in a dormant state until triggered by an event that requires them to perform their function.
Narrowband IoT (NB-IoT) is a low power wide area (LPWA) network protocol that has been developed to support M2M communications between these types of devices, recognising that they have low data transfer rates and don’t continually transmit and receive data.
Due to the specific characteristics of M2M communications, NB-IoT must address a number of challenges, including:
This article describes a new procedure developed by CEVA to improve the initial synchronisation process, when the device wakes up, thereby reducing the NPSS detection latency.
Synchronisation, NPSS and NSSS
When an event occurs which requires an IoT device to exit sleep mode, the device must follow a defined sequence:
The NPSS and Narrowband Secondary Synchronisation Signals (NSSS) are instrumental to the synchronisation process. These signals are repeated every one and two radio frames and the device must continue to receive data from the base station until the NPSS signal is detected.
Whilst NB-IoT is heavily based on the LTE protocol, the synchronisation signals have been completely redesigned to enable deployment inside and adjacent to LTE carriers.
Figure 1 shows how the synchronisation signals are positioned in the radio frame. The signals use 180 kHz bandwidth, with the NPSS located in the 6th sub-frame and the NSSS in the 10th sub-frame of every even radio frame.
Figure 1 Radio frame synchronisation signals
The NPSS, illustrated in Figure 2, is transmitted every 10 msec and is constructed of the last 11 OFDM symbols of the sub-frame and uses the first 11 subcarriers of the allocated Physical Resource Block (PRB).
Figure 2 NPSS structure (frequency domain)
In order to improve efficiency and performance of the synchronisation process, a pseudo-random code is used to generate the OFDM symbols, as depicted in Figure 3. This code is predetermined and enables the NPSS to be known to the device. However, the significant uncertainty in timing and frequency of the transmitted signal results in NPSS detection having a high processing overhead for the device.
Figure 3 NPSS structure (time domain)
The NSSS has a similar structure to the NPSS, as shown in Figure 4 but uses all 12 PRB sub-carriers of the PRB and consists of 132 frequency domain elements, 12 sub-carriers for every 11 OFDM symbol.
The binary code used to generate the NSSS enables it to be predetermined and known by the IoT device and is used to determine the cell ID and radio frame index. The IoT device can only determine these parameters after completion of time and frequency synchronisation.
Upon completion of time and frequency estimation and detection of the cell ID and radio frame boundary, the device can begin to decode the control and data channels.
NB-IoT synchronisation techniques
The primary synchronisation procedure, described above, actually consists of two stages – the coarse stage and the fine stage, as illustrated in Figure 4.
Coarse stage synchronisation is done at the lower frequency of 240 kHz, in order to reduce processing load, and provides estimations of the timing and frequency of the received signal.
The fine stage provides refined timing and frequency estimations with lower errors than the parameters estimated in the coarse stage. This happens by sampling the signal at 1.92 MHz around the coarse time estimation and then shifting the signal’s frequency to baseband, based on the coarse frequency estimation.
Following estimation of timing and frequency, the IoT device can carry on to detect the cell ID and frame boundary, based on NSSS detection.
The remainder of this article looks at two known estimation methods for the coarse stage of primary synchronisation, before considering the new method designed by CEVA. Finally, latency performances for each of the three methods are simulated and compared.
Figure 4 Synchronisation process
[Continue reading on EDN US: Cross-correlation synchronisation]
Emmanuel Gresset is a business development director in CEVA Wireless BU.