Communication Basics #34 Worked Example: STM-1 + Mixed 53/77-byte Payload — 85.19% Real Data

Instructor: Dr. Süleyman Burak ÇELİK

The third and final video in our STM-1 filling trilogy. Pure ATM gave us 83.17% efficiency. Pure 576-byte IP gave us 90.87%. But real networks rarely carry just one kind of traffic. Today we mix two cell sizes inside the same STM-1 frame — 53-byte ATM cells alternating with 77-byte cells — and watch the answer land cleanly between the two: 85.19% efficiency, 132.48 Mbps, the weighted-average law in action. Two cell types side by side. Cell A is the familiar ATM cell — 5-byte header plus 1-byte AAL plus 47 bytes of user data, for 53 bytes total. Cell-level efficiency 47/53 ≈ 88.7%. Cell B is a 77-byte cell — 5-byte header plus 72 bytes of user data, no separate AAL byte (the larger payload absorbs that role). Cell-level efficiency 72/77 ≈ 93.5%, about 5 points better. Same 5-byte header, but spread over a 50% larger payload — exactly the same lesson from the IP video, just applied at smaller scale. But voice still needs small fixed cells for predictable latency. So networks mix. The question is how the math works when the payload is heterogeneous. The packing arithmetic. The alternating pair is one ATM plus one large cell, 53 + 77 = 130 bytes per pair. 2 322 / 130 = 17.86, round down to 17 complete pairs. That occupies 17 × 130 = 2 210 bytes, leaving 112 bytes of tail. The squeeze: can we fit one more ATM cell? 53 ≤ 112, yes. Can we then fit one more large cell? 77 ≤ 59, no. Final: 18 ATM cells plus 17 large cells per STM-1 frame, with 59 bytes of tail wasted. User data: 18 × 47 + 17 × 72 = 846 + 1 224 = 2 070 bytes per frame, or 16 560 user bits. Efficiency: 16 560 / 19 440 ≈ 85.19%. Data rate: 16 560 / 125 µs = 132.48 Mbps. Mixed payload sits about two points above pure ATM, five points below pure IP — exactly where a weighted blend should land. Three takeaways from the trilogy. (1) STM-1 is fixed scaffolding — 9 × 270 bytes, 19 440 bits, 8 000 frames per second, 155.52 Mbps line rate. The frame is the constant; the payload is the variable. (2) Efficiency = whatever you fill the payload with. Small fixed cells like ATM trade efficiency for predictable latency. Big variable packets like IP win efficiency at the cost of bursty timing. Mixed payloads land in between, scaled by the user-byte weights of each cell type. (3) This is real-network math. No metro carrier link runs pure ATM or pure IP — voice, video, data, signaling all share the same frame. The math we just walked through is exactly what a network engineer runs when sizing a link. Trilogy closes here. Topics covered: - two cell types side by side (53B ATM at 88.7% vs 77B cell at 93.5%) - the alternating pair: 1 ATM + 1 large = 130 bytes, 91.5% pair efficiency - pairs per frame: floor(2 322 / 130) = 17 pairs, 112-byte tail remaining - tail squeeze: +1 ATM (53 ≤ 112) but no more large (77 is greater than 59) → 18 ATM + 17 large - user data per frame: 18×47 + 17×72 = 2 070 bytes = 16 560 bits - mixed efficiency 85.19% — between pure ATM (83.17%) and pure IP (90.87%) - weighted-average law: η_mix = sum of per-cell efficiencies weighted by user-byte share - user data rate 132.48 Mbps — the price of mixing vs the benefit of operational flexibility Sources: ECE 271 ExQ52. Closes the STM-1 trilogy: d32 (pure ATM) → d33 (pure IP) → d34 (mixed). Part of the AcEdumy Communication Basics series.