Telegram Splitting Ultra Narrow Band

Discussion

• What are the radio characteristics of TS-UNB?

  

  TSMA scheme of a radio frame. Source: ETSI 2018a, fig. 6-17.

  TS-UNB is defined in license-free bands 868 MHz (Europe) and 915 MHz (US). Channel bandwidth is based on the TSMA mode: Narrow (25 kHz), Standard (100 kHz), and Wide (750 kHz). The corresponding carrier spacing is about 397 kHz, 2.4 MHz, and 28.6 MHz.

  Radio-burst duration is based on the protocol mode: UL-ULP (15.14 ms), UL-ER (90.74 ms), and DL-TS-ULP (11.76-12.43 ms), for Ultra Low Power (ULP), Extended Reach (ER) and Time Splitting (TS). A radio burst is followed by radio idle time, thus limiting the duty cycle as per the TSMA pattern.

  In DL-TS, a message is split and sent across many radio bursts. A set of radio bursts carrying one message is called a radio frame. Bursts are spread across 24 carriers (C0-C23). Carrier C24 is reserved for the optional Sync-Burst. Radio frame consists of core frame and extension frame:

  • Uplink: Core (24 bursts) plus additional extension frame (one radio burst for every extra byte) if PHY payload exceeds 186 bits.
  • Downlink: Core (9 bursts repeated to fill 18 bursts) plus optional extension frame (18 bursts per block, multiple blocks).
• What are the operation modes in TS-UNB?

  

  TS-UNB modes. Source: ETSI 2018a, table 6-1.

  TS-UNB has a few operation modes. An LTN system must implement one subset. Downlink modulation can be GMSK (with telegram splitting) or GFSK (with single burst) whereas uplink modulation is always GMSK (with telegram splitting). Variable MAC mode allows implementers to define custom MAC/link layers or use Wireless M-Bus.

  The standard defines TS-UNB Profiles, which are sets of modes to enable interoperability between endpoints and base stations. The following profiles are defined:

  • EU0: For Europe. Telegram splitting. Single channel. 100 kHz bandwidth. Standard TSMA mode.
  • EU1: Similar to EU0 but dual channel.
  • EU2: Similar to EU1 but 750 kHz bandwidth and Wide TSMA mode.
  • US0: Similar to EU2 for U.S. bands.

  In addition, device classes are defined. Class Z devices are uplink only. Class A devices are capable of bidirectional communication, with downlink transmission at the base station triggered by an uplink reception. Class A devices negotiate at the link layer some operation modes. The configuration is fixed in the standard for Class Z devices.

• How is a TSMA pattern selected for the next transmission?

  

  UPG2: 24 carriers defined for each of 8 patterns. Source: ETSI 2018a, table 6-51.

  Within a radio frame (core + extension), the time-frequency allocation for transmission is called a TSMA Pattern. A group of patterns is called a TSMA Pattern Group. Patterns and their groups are fixed in the standard.

  Uplink has three groups: UPG1 (8 patterns), UPG2 (8 patterns) and UPG3 (1 pattern). UPG1 is for a single transmission. UPG2 is for initial and repeat transmissions of a radio frame. UPG3 is to achieve low latency. Downlink has only one group called DPG (8 patterns).

  For UPG1 and UPG2, the endpoint shall use the patterns in the sequence (1,2,3,4,1,2,3,4,5,1,2,3,4,5,6) and repeat this sequence. Pattern 6 is used only for attach process, pattern 7 for high priority messages and pattern 8 is reserved. For extension frame, the pattern is selected pseudo-randomly. In downlink, the pattern is selected pseudo-randomly based on the Header CRC field. The same pattern is used for both core and extension frames.

• What's frame repetition in TS-UNB?

  For better performance due to redundancy, a radio frame may be repeated. This is possible in both uplink and downlink. In uplink, the initial and repeated radio frames shall have identical data. Applicable pattern is UPG2. In downlink, extended frame is not repeated. Downlink repeated frame uses the second half of the TSMA pattern.

  Header CRC and Payload CRC fields of the uplink PHY payload are calculated and set by the endpoint. These fields are available to the base station. They determine the time offset and frequency offset for the repeat transmission. Time offset is zero for initial transmission. Frequency offset is zero for single channel transmission. Frequency offset is always the same for initial and repeat transmissions.

  The use of repetition itself is configured at the link layer during the attach procedure.

• What are the PHY layer functions in TS-UNB?

  

  Block diagram of PHY waveform generation. Source: ETSI 2018a, fig. 6-12.

  TS-UNB doesn't define any time synchronization across the network. A radio frame starts when the first burst is sent by an endpoint. This starting point is randomly chosen by the endpoint.

  In the downlink, block slicing happens. Each Block PSDU is limited to 24 bytes and gets its own CRC. This improves error detection. Moreover, if a Sync-burst is used on the core frame, it's also used on each extension frame, thus aiding the receiver.

  Long sequences of zeros or ones can cause problems at the receiver. This is prevented by doing data whitening using the PN9 pseudo-random sequence generator.

  For error correction, convolutional code at rate 1/3 is used.

  Finally, PHY payload is split into many sub-packets. This is done via interleaving, which assigns the bits to each sub-packet and then places the bits in the correct order within each burst. In the uplink, CRC fields and PSI appear in core frame. In the case of DL-SB, PHY payload is sent in a single burst without splitting.

• What's the format of a TS-UNB radio burst?

  

  TS-UNB radio bursts. Source: Adapted from ETSI 2018a.

  Uplink PHY PDUs are sent out as multiple radio bursts over the radio frame. Each burst has 12-bit data, 12-bit pilot sequence (PS) and another 12-bit data. Pilot sequences differ for the core frame and the extension frame.

  In downlink, the core frame is to used to acknowledge the uplink endpoint transmission. Thus, it may be used on its when there's nothing to send on the downlink. Extension frame carries the PHY PDUs in blocks. Each block has a slice of the PPDU called Block PSDU of at most 24 bytes. For DL-TS, each burst has at the minimum 8-bit PS, 12-bit data and another 8-bit PS. A burst in an extension frame can contain two more data fields of variable length.

  A downlink Sync-burst may precede the core frame and therefore each block. A core frame Sync-burst has 5-bit PS, 12-bit data, and another 4-bit PS. Sync-bursts in the extended frame have variable number of pilot symbols depending on symbol availability.

• What are the MAC layer functions in TS-UNB?

  

  Uplink/downlink scheduling. Source: ETSI 2018a, fig. 6-11.

  MAC takes care of scheduling. While uplink transmission may start anytime, downlink transmission starts 16,384 symbols after receiving the last uplink radio burst. Time offsets are based on midpoints of the pilot sequence within each burst.

  Response flag in the MAC header announces the downlink window. Both uplink and downlink MAC headers have a response flag. If set, the receiver is expected to acknowledge the reception. However, base station may ignore this flag due to its limited transmission duty cycle. When priority flag is set in the downlink, endpoint should acknowledge within a time window.

  MAC does endpoint addressing. Downlink addressing is implicit in the timing and CMAC. Uplink addressing is via a 16-bit short address (received during attach procedure) or a globally unique IEEE EUI64 address.

  MAC does encryption of its payload. Except for the attach packet request, all others are encrypted. AES128 algorithm is used. The attach procedure provides information to determine the 128-bit network key based on pre-shared private key. In addition, MAC generates and includes a 32-bit CMAC for the integrity and authenticity of each packet.

• What are the link layer functions in TS-UNB?

  

  Endpoint attach procedure for TS-UNB Class Z and A devices. Source: Adapted from ETSI 2018a.

  Link layer establishes and maintains a connection between the endpoint and the base station. It has a number of control segments to achieve this: Attach Request/Accept, Detach Request/Accept, DLRX-Status Query/Response, Link Adaptation Request/Confirm, and user-specific control segments. Attach requests are sent by endpoint. Detach requests can be sent by either end. Detach disconnects the endpoint from the network. Since class Z devices are unidirectional, they're pre-attached during manufacturing.

  During attach request/accept and link adaptation request, endpoint and base station agree on the capabilities via the Endpoint Info field: single/dual channel, use of repetition, DL interblock distance, etc. A default configuration is used for the attach procedure.

  Apart from its control functions, the link layer also relays to MAC data payload coming from the network layer above.

• What's the TS-UNB PDU structure at different layers?

  

  TS-UNB uplink and downlink LPDU, MPDU and PPDU formats. Source: Adapted from ETSI 2018a.

  At the link layer, the same format is used both in uplink and downlink. LPDU consists of control payload and data payload. Control payload can contain one or more control segments. Each control segment includes a 2-bit header to indicate the type of control message such as Attach, Detach, Link Adaptation, etc.

  At MAC, fixed mode and variable mode have different formats. MPDU in variable mode has the same format in uplink and downlink. The chosen mode is indicated by the MMode field in PPDU. SIGN field is a Cipher-based Message Authentication Code (CMAC).

  At PHY, uplink PSDU is 20-255 bytes. PSDU less than 20 bytes is zero-padded and padding is ignored for CRC calculation. Payload CRC is based on MPDU and MMode. Header CRC is based on Payload CRC and PSI. Packet Size Indicator (PSI) gives the actual size before padding. In the downlink, the same PPDU format is used for both DL-SB and DL-TS. Downlink PSDU is 7-250 bytes.

Milestones

Sep
2011

Performance analysis of telegram splitting. Source: Bernhard and Kilian 2014, fig. 9.

Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. files a German patent that introduces the concept of telegram splitting. Their analysis shows that the probability of losing a telegram decreases when it's split into more packets. Curve 170 in the figure shows packet error probability while other curves show telegram error probability. The patent is subsequently published in the U.S. as US20140176341 (June 2014).

Sep
2014

ETSI publishes an initial set of documents that specify the architecture, protocols, interfaces and use cases of Low Throughput Networks (LTNs). At this point, there's no explicit mention of telegram splitting.

Dec
2015

Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. apply for a German trademark on MIOTY. International and U.K. trademarks are filed in May 2016. Subsequently, these applications are approved and registered. MIOTY becomes the commercial name for Fraunhofer's LPWAN solution that adopts TS-UNB as the underlying protocol. Sisvel International manages the licensing of MIOTY and this is announced in April 2020.

Jun
2018

ETSI publishes two documents: TS 103 357: Protocols for radio interface A and TS 103 358: LTN Architecture. For the radio interface A, the standard defines three different families. TS-UNB is one of them, thus becoming an industry standard.

References

1. Bernhard, J. and G. Kilian. 2013. "DE102011082098 - Batteriebetriebene stationäre Sensoranordnung mit unidirektionaler Datenübertragung." German patent, Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V., March 7. Filed 2011-09-02. Accessed 2022-06-02.
2. Bernhard, J. and G. Kilian. 2014. "US 2014/0176341 A1 - Battery-operated stationary sensor arrangement with unidirectional data transmission." U.S. patent, Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V., June 26. Filed 2014-02-26. Accessed 2022-06-02.
3. Blackman, James. 2019. "What is MIOTY? All about telegram splitting and LoRaWAN bashing." Enterprise IoT Insights, RCRWireless, September 26. Accessed 2022-05-31.
4. Business Wire. 2020. " Sisvel Announces the Launch of its MIOTY LPWAN Licensing Program." Business Wire, April 14. Accessed 2022-06-02.
5. Christiansen, Grant. 2010. "Data Whitening and Random TX Mode." Design Note DN509, Texas Instruments. Accessed 2022-06-03.
6. ETSI. 2014a. "GS LTN 001: Low Throughput Networks (LTN); Use Cases for Low Throughput Networks." V1.1.1, September. Accessed 2021-06-02.
7. ETSI. 2014b. "GS LTN 002: Low Throughput Networks (LTN); Functional Architecture." V1.1.1, September. Accessed 2021-06-02.
8. ETSI. 2014c. "GS LTN 003: Low Throughput Networks (LTN); Protocols and Interfaces." V1.1.1, September. Accessed 2021-06-02.
9. ETSI. 2018a. "TS 103 357: Short Range Devices; Low Throughput Networks (LTN); Protocols for radio interface A." V1.1.1, June. Accessed 2022-06-31.
10. ETSI. 2018b. "TS 103 358: Short range devices; Low Throughput Networks (LTN) Architecture; LTN Architecture." V1.1.1, June. Accessed 2022-06-31.
11. EUIPO. 2022. "Search for MIOTY." TMview, EUIPO. Accessed 2022-06-02.
12. Radiocrafts. 2022. "ETSI Compliant Communication Solution For LPWANS (ETSI TS 103 357)." Radiocrafts, February 25. Accessed 2022-05-31.
13. Sisvel. 2022. "About MIOTY Licensing Platform." Sisvel International. Accessed 2022-06-02.

Further Reading

1. ETSI. 2014a. "GS LTN 001: Low Throughput Networks (LTN); Use Cases for Low Throughput Networks." V1.1.1, September. Accessed 2021-06-02.
2. Bernhard, J. and G. Kilian. 2014. "US 2014/0176341 A1 - Battery-operated stationary sensor arrangement with unidirectional data transmission." U.S. patent, Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V., June 26. Filed 2014-02-26. Accessed 2022-06-02.
3. Radiocrafts. 2022. "ETSI Compliant Communication Solution For LPWANS (ETSI TS 103 357)." Radiocrafts, February 25. Accessed 2022-05-31.

Article Stats

Author-wise Stats for Article Edits

Author

No. of Edits

No. of Chats

DevCoins

arvindpdmn

4

0

1504

DevCoins due to articles, chats, their likes and article hits are included.

Cite As