# SMS Fundamentals & Encoding
URL: /docs/guides/sms-fundamentals

> Telecommunication network architecture, GSM-7 vs. UCS-2 character sets, segmentation math, and automatic phone normalization.

Short Message Service (SMS) is a store-and-forward message. This page covers how a message moves from your app to a handset, how encoding changes the segment count, and how phone numbers are cleaned on the way in.

---

## 1. Path of a message [#1-path-of-a-message]

![Your app, SMS Pro, the carrier, and the handset](/images/telco-network-architecture.svg)

SMS Pro accepts the API call. The carrier delivers the SMS. The handset result comes back later as a delivery status.

---

## 2. Character Encodings: GSM-7 vs. UCS-2 [#2-character-encodings-gsm-7-vs-ucs-2]

Every character in an SMS payload is encoded as either **GSM 7-bit default alphabet** or **UCS-2 (16-bit Unicode)**. This distinction directly controls message segment capacity and billing units.

### GSM 7-bit Encoding (Standard) [#gsm-7-bit-encoding-standard]
The GSM 03.38 standard defines 128 basic characters (A-Z, a-z, 0-9, standard punctuation, and select Greek symbols). Because each character requires only 7 bits of data, a standard 140-byte telecommunication signaling packet holds up to **160 characters**.

- **Single Segment Maximum**: **160 characters**.
- **Concatenated Segment Maximum**: **153 characters** (7 bytes are consumed by the User Data Header for reassembly).

### UCS-2 Encoding (Unicode) [#ucs-2-encoding-unicode]
If an SMS payload contains even a single character outside the GSM-7 table (such as an emoji 👋, accented character, Arabic, Chinese, or specialized punctuation like curly quotes `“`), the entire message must be encoded using **UCS-2**.

Each character in UCS-2 requires 16 bits (2 bytes). This drastically reduces the capacity of a 140-byte signaling packet:

- **Single Segment Maximum**: **70 characters**.
- **Concatenated Segment Maximum**: **67 characters** (6 bytes are consumed by the User Data Header).

| Metric | GSM 7-bit Encoding | UCS-2 Unicode Encoding |
| :--- | :--- | :--- |
| **Bits Per Character** | 7 bits | 16 bits (2 bytes) |
| **Single Message Capacity** | **160 characters** | **70 characters** |
| **Multi-Part Segment Capacity**| **153 characters / segment** | **67 characters / segment** |
| **Trigger Characters** | Latin letters, numbers, standard symbols | Emojis (🚀), smart quotes (“”), non-Latin alphabets |
| **Billing Impact** | 1 unit per 160 chars | 1 unit per 70 chars (**~2.3x more units**) |

<Callout type="warn" title="The Single-Emoji Multiplier">
A 150-character plain text message is billed as **1 SMS unit**. Adding a single emoji converts the message to UCS-2, splitting the 151 characters into **3 concatenated segments** (67 + 67 + 17 characters), tripling your unit cost.
</Callout>

---

## 3. SMS Segmentation & Concatenation [#3-sms-segmentation--concatenation]

When a message exceeds the single-segment threshold, the SMS Pro gateway divides the text into concatenated segments. Each segment includes a 6-byte **User Data Header (UDH)**:

- **UDH Contents**: Information Element Identifier (IEI), CSMS reference number, total segment count (N), and current segment index (n).
- **Handset Reassembly**: The receiving mobile device buffers the incoming segments, orders them using the UDH sequence numbers, and displays them as a single seamless text to the user.

### Segmentation Calculation Example (GSM-7) [#segmentation-calculation-example-gsm-7]

```text
segments = length <= 160 ? 1 : Math.ceil(length / 153)
```

- Length = 160 characters $\to$ **1 segment** (160 used).
- Length = 161 characters $\to$ **2 segments** (153 + 8 used).
- Length = 306 characters $\to$ **2 segments** (153 + 153 used).
- Length = 307 characters $\to$ **3 segments** (153 + 153 + 1 used).

---

## 4. Phone Number Normalization & E.164 [#4-phone-number-normalization--e164]

SMS Pro cleans numbers on ingest:

![Phone number normalization](/images/phone-normalization-flow.svg)

### Supported Input Formats [#supported-input-formats]
- **Local with leading zero** (`0241234567`): 10 digits; leading `0` is replaced with national dial code `233`.
- **National without zero** (`241234567`): 9 digits; prepends national dial code `233`.
- **International without plus** (`233241234567`): 12 digits; validated and preserved.
- **E.164 with plus** (`+233241234567`): The leading `+` is removed.
- **Punctuation & Spaces** (`+233 (24) 123-4567`): All non-digit characters are automatically stripped.

### Automatic Deduplication [#automatic-deduplication]
If an API request or campaign file lists both `0241234567` and `+233241234567`, the engine detects identical normalized MSISDNs and deduplicates the recipient to prevent duplicate message delivery and billing.

---

## 5. Architectural Best Practices [#5-architectural-best-practices]

1. **Sanitize Dynamic Template Inputs**: If inserting user names or external database strings into SMS templates, strip invisible non-GSM characters (e.g. zero-width spaces or smart quotes) to prevent unintended UCS-2 conversion.
2. **Pre-Calculate Segment Counts**: Check the `segmentCount` in the `202 Accepted` response to monitor billing consumption in real time.
3. **Use Shortlinks for Long URLs**: URLs consume massive character counts. Use SMS Pro Branded Shortlinks to condense links to under 20 characters while gaining click analytics.

---

## Related Documentation [#related-documentation]
- 📖 [Message Lifecycle & Routing](/docs/guides/message-lifecycle-and-routing)
- ⚡ [Recipe: Idempotent SMS Dispatch](/docs/recipes/idempotent-sms-dispatch)
- 🔌 [API Reference: Send SMS](/docs/api-reference/transactional-sms/sendSms)