Putting an IPv6 stack inside a battery-powered water meter is not hard because of compute. It is hard because of headers. A standard IPv6 header is 40 bytes; add UDP and you are at 48 bytes of protocol overhead before a single byte of application data — perilously close to a full LoRaWAN frame, and far beyond the 15-byte uplink cap of CN470 at DR0.

That is the seam between the IP world and LPWAN. SCHC (Static Context Header Compression, pronounced "chic") is the IETF standard built for this seam — and currently the only standardized way to carry full end-to-end IPv6 over LoRaWAN. This article unpacks the core mechanism, the fragmentation design, the ecosystem today, and the more practical question: which projects need it, and which do not.

1. What SCHC Is: Three Documents, One Protocol Family

SCHC was developed by the IETF LPWAN working group and consists of three documents:

The same mechanism has since been adapted to IEEE 802.15.4 and Sigfox (RFC 9442), but this article stays on the LoRaWAN line.

One common misconception to clear up first: SCHC is not a LoRa Alliance specification — it is an IETF RFC. The LoRaWAN adaptation only defines how SCHC messages are carried in LoRaWAN frames, plus parameters like ports and timers. This is also why SCHC sits alongside Relay and FUOTA in the LoRa Alliance certification framework: protocol certification for these three features is only available at ATL laboratories (as covered in our certification article — mainland China currently has no ATL, so samples must travel abroad).

2. The Core Idea: Never Transmit the Context, Only the Delta

The hard problem in header compression has always been: how do both ends agree on what was compressed? Classical schemes such as RoHC (used in cellular) learn as they go — the compressor observes traffic, builds context dynamically, and both ends continuously synchronize state. That learning phase depends on sustained traffic, and LPWAN devices have none: four uplinks a day, sleep in between. The context expires before it is established.

SCHC inverts this: the context never crosses the network — both ends are pre-provisioned with the same rule set. That is what "static" means. Rules are installed at manufacturing or commissioning and do not change during transmission; no synchronization messages exist.

The design exploits exactly the three properties of LPWAN traffic: it is predictable (a meter's uplink always carries the same fields), routing-free (star topology), and tiny (every byte saved is real airtime and real battery).

How compression works in practice:

  1. The sender splits the header into fields and matches them against the rules;
  2. Each rule declares an expected value per field — the device's IPv6 prefix is always the operator-assigned one, the UDP port is always the agreed one;
  3. Fields that match their expected values are not transmitted at all — the packet carries only the rule identifier (RuleID); only deviations travel as residue bits;
  4. The receiver looks up the rule by RuleID and reconstructs the full header.

The result: a 48-byte IPv6/UDP header typically compresses to a 1-byte RuleID plus a few bits of residue. If a header matches no rule, the protocol defines a "no-compression" rule as a fallback — FPort 22 in the LoRaWAN adaptation. The packet always gets through; it just saves nothing.

3. RuleID and Ports: What the LoRaWAN Adaptation Specifies

RFC 9011 carries SCHC messages inside LoRaWAN frames in a direct way: the RuleID is encoded in the LoRaWAN FPort field (8 bits); the FPort plus the payload form the complete SCHC message.

Recommended port assignments in the adaptation: FPort 20 for uplink fragmentation, FPort 21 for downlink fragmentation, FPort 22 for the no-compression fallback; remaining ports are available for compression rules. Applications can keep using their own ports for non-SCHC traffic — the two coexist.

One privacy detail worth noticing: the recommended algorithm for generating the device's IPv6 interface identifier (IID) in SCHC over LoRaWAN applies AES-CMAC with the AppSKey to the DevEUI and takes the first 64 bits. Because AppSKey changes on every rejoin, the device's IPv6 address changes with each network join — protection against long-term address tracking, built into the protocol's recommended algorithm.

4. Fragmentation: When It Still Does Not Fit

Compression solves "the header is too big"; fragmentation solves "the packet is too big." SCHC fragmentation has three modes:

The LoRaWAN adaptation pins concrete uplink parameters: a 2-byte fragment header (FPort plus one byte), 4 windows of 63 tiles with 10 bytes of payload per tile — a maximum reassembled datagram of 2,520 bytes. The recommended retransmission and inactivity timers default to 12 hours. That number captures SCHC's design philosophy: it assumes a slow, lossy link and devices that are mostly asleep, and prefers hour-scale waiting over device-side power drain.

The document adds one recommendation specifically for battery devices: acknowledge at the end of each window instead of at the end of the whole datagram — otherwise a device that has lost coverage keeps transmitting uselessly, burning battery for nothing.

5. Ecosystem and Use Cases: Metering Is the Main Battleground

The commercial ecosystem currently centers on France's Acklio, which provides the reference implementation. Its claimed deployment benefits: 40–90% less data traffic, 30–70% less energy consumed, and up to 7× more devices per cell (vendor figures; actual gains depend on traffic patterns).

The highest-stakes use case is metering. DLMS/COSEM is the dominant application-layer standard in electricity, gas and water metering, and it natively runs on an IP stack. SCHC is positioned as the standard adaptation layer for DLMS over LoRaWAN — for legacy DLMS meters moving onto LoRaWAN, SCHC is the mandatory stop on the standardized route. The same logic extends to Modbus over LoRaWAN, LwM2M device management, and other IP-family protocols going over the air.

This is where SCHC intersects our readers' business most directly. Legacy wired meters and Modbus devices going wireless is one of the main growth engines of LoRaWAN — and the larger the meter vendor, the less willing it is to rewrite a private binary protocol per customer. A standard IP stack plus SCHC is the "don't touch the meter, adapt the network" route.

6. Do You Need SCHC? Three Tests

Signs you need it: your devices already run an IP stack (IPv6/LwM2M/DLMS) and you do not want to rewrite the protocol for LoRaWAN; you need end-to-end IP reachability (devices addressable from the internet); multiple vendors' devices must integrate under one standard instead of negotiating private payload formats.

Signs you do not: your devices already speak a custom binary payload of a dozen-odd bytes — there is nothing for compression to squeeze, and on that line, thing-model decoder scripts (the approach from our article 21) are the more direct tool; your datagrams are small and stable, never exceeding the frame size, so fragmentation buys nothing; your team has no IPv6 operations capability, so end-to-end addressing costs more than it returns.

A pragmatic positioning: SCHC solves "IP-world protocols must pass through the LoRaWAN keyhole." If your devices speak a binary dialect and the platform does thing-model translation (ThinkLink's default route), that keyhole is not on your path. If your devices speak standard IP protocols and resist modification, SCHC is the standardized way through. ThinkLink's current stated compatibility is LoRaWAN 1.0.2 / 1.0.3 / 1.0.4; SCHC is an IETF-side adaptation layer, and its platform-side support timeline should be confirmed with the vendor directly — when evaluating such projects, verify the NS-side SCHC gateway capability before touching the device side.

7. Costs and Risks

SCHC is not free. The device side maintains rule storage plus compression/fragmentation state machines (memory footprint is smaller than the 6LoWPAN route, but real), and rule configuration is an engineering task of its own — both ends must agree field by field, and a misconfigured rule fails silently at decompression, which means troubleshooting at the SCHC layer, not the LoRaWAN layer. Rule distribution is explicitly out of scope in the standard; real deployments use vendor tooling or commissioning-time provisioning, and this part carries no cross-vendor interoperability guarantee.

Add the certification constraint: SCHC protocol certification is ATL-only, with the same mainland-China scheduling problem discussed for Relay. Sum it up and SCHC fits scenarios where the standardization benefit exceeds the integration cost — typically standards-driven industries such as metering and utilities, not a single project with a few dozen devices.

Closing Thoughts

Relay solves "the signal cannot get there." SCHC solves "the protocol does not fit." Both treat LPWAN constraints as design inputs rather than obstacles: Relay trades one hop and 16 devices for operability; SCHC trades static rules for the elimination of dynamic synchronization, compressing a 48-byte IP header down to 1 byte.

For most projects, SCHC will remain at the "good to know it exists" level. But for teams doing metering, DLMS retrofits or IP-protocol migration, it is the standard you cannot route around. Knowing its boundaries in advance — rule engineering, silent failures, ATL-only certification — is far cheaper than discovering them mid-project.

(End)