Loading learning content...
Now that we understand carrier sensing—the ability to detect whether the channel is busy—we face a critical design question: What should a station do when it has data to send?
The simplest and most aggressive answer is 1-persistent CSMA: sense the channel, and if it's idle, transmit immediately with probability 1 (hence the name). If the channel is busy, wait until it becomes idle, then transmit immediately.
This approach maximizes the chance of successful transmission when the channel is truly free, but it comes with a significant drawback: if multiple stations are waiting for a busy channel to become idle, they will ALL transmit simultaneously the moment the channel clears, guaranteeing a collision.
By the end of this page, you will understand: (1) The precise algorithm of 1-persistent CSMA, (2) Why 'persistence' refers to behavior when the channel is busy, (3) The collision scenarios specific to 1-persistent CSMA, (4) Mathematical analysis of throughput, and (5) When 1-persistent CSMA is appropriate and when it fails.
Before diving into 1-persistent CSMA, let's clarify what persistence means in this context. It's a common source of confusion.
Persistence refers to what a station does when it finds the channel BUSY—not what it does when the channel is idle.
When a station senses a busy channel, it has several options:
The term 1-persistent means the station:
The '1' in '1-persistent' is the probability of transmission when idle is detected. We'll see later that p-persistent CSMA transmits with probability p < 1.
| Strategy | When Channel Busy | When Channel Idle | Transmission Probability |
|---|---|---|---|
| 1-Persistent | Keep sensing continuously | Transmit immediately | p = 1 (always) |
| Non-Persistent | Wait random time, then re-sense | Transmit immediately | p = 1 (always) |
| p-Persistent | Keep sensing continuously | Transmit with prob. p; defer with prob. (1-p) | p < 1 (probabilistic) |
1-persistent stations 'camp' on a busy channel—they refuse to back off and constantly monitor, ready to pounce the instant the channel becomes idle. This greedy behavior leads to high utilization when traffic is light, but guaranteed collisions when multiple stations are waiting.
Let's formalize the 1-persistent CSMA algorithm step by step. This precise specification eliminates ambiguity about station behavior.
Key behavioral aspects:
Immediate transmission when idle: There is no random delay when the channel is sensed idle. The station transmits at the first opportunity.
Persistent monitoring when busy: The station does not give up and check later. It continuously samples the channel state, ready to transmit the instant the channel becomes free.
Greedy seizure: The station seizes the channel at the earliest possible moment. This minimizes latency for the individual station but creates contention with other waiting stations.
The aggressive nature of 1-persistent CSMA creates a specific, predictable collision scenario. Let's trace through a concrete example.
Scenario: Four stations (A, B, C, D) share a channel. Station X is currently transmitting. During X's transmission, stations A, B, and C all generate frames they want to send.
Timeline:
| Time | Event | Station A | Station B | Station C | Channel |
|---|---|---|---|---|---|
| t=0 | X starts transmitting | — | — | — | Busy |
| t=10 | A generates frame | Senses busy → Waits | — | — | Busy |
| t=20 | B generates frame | Waiting | Senses busy → Waits | — | Busy |
| t=50 | C generates frame | Waiting | Waiting | Senses busy → Waits | Busy |
| t=100 | X finishes | Detects idle → Transmits | Detects idle → Transmits | Detects idle → Transmits | Idle→COLLISION |
The result: Three stations were waiting. The moment the channel became idle, all three transmitted simultaneously. Guaranteed collision.
In 1-persistent CSMA, if N stations are waiting for a busy channel to become idle, ALL N will transmit simultaneously when the channel clears. Under high load, this creates cascading collisions: the collision causes retries, more stations wait, the next transmission creates an even bigger collision, and so on.
Two types of collisions in 1-persistent CSMA:
Type 1: Propagation Delay Collisions These occur even with a single transmitter. Station A starts transmitting; Station B senses idle (hasn't received A's signal yet) and also transmits. This is inherent to all CSMA.
Type 2: Waiting Station Collisions These are specific to 1-persistent behavior. Multiple stations wait for a busy channel; they all transmit simultaneously when it becomes idle. These collisions are avoidable with different persistence strategies.
Type 2 collisions become dominant under high load, making 1-persistent CSMA perform poorly when the network is congested.
Let's analyze the throughput of 1-persistent CSMA mathematically. We'll use the standard model with:
Assumptions:
Throughput derivation (simplified):
For 1-persistent CSMA, the throughput S as a function of offered load G is given by:
$$S = \frac{G \cdot e^{-G(1+2a)} \cdot (1 + G + aG(1 + G + aG/2))}{G(1 + 2a) - (1 - e^{-aG}) + (1 + aG) \cdot e^{-G(1+a)}}$$
This complex expression simplifies in limiting cases:
When a → 0 (negligible propagation delay): $$S = \frac{G \cdot e^{-G} \cdot (1 + G)}{G + e^{-G}}$$
When a is small (typical LAN): The maximum throughput approaches: $$S_{max} \approx \frac{1}{1 + 3.44a}$$
For a = 0.01 (typical Ethernet): S_max ≈ 97%
| Offered Load (G) | Throughput (S) | Efficiency | Wasted Capacity |
|---|---|---|---|
| 0.1 | 0.095 | 95% | 5% |
| 0.5 | 0.42 | 84% | 16% |
| 1.0 | 0.53 | 53% | 47% |
| 2.0 | 0.45 | 23% | 77% |
| 5.0 | 0.24 | 5% | 95% |
| 10.0 | 0.12 | 1% | 99% |
Notice that beyond G ≈ 1, throughput actually DECREASES as offered load increases! This is the 'overload collapse' phenomenon. More traffic leads to more collisions, which requires more retransmissions, which creates even more traffic. The network becomes congested and throughput plummets.
Comparison with ALOHA:
| Protocol | Max Throughput | Offered Load at Max |
|---|---|---|
| Pure ALOHA | 18.4% | G = 0.5 |
| Slotted ALOHA | 36.8% | G = 1.0 |
| 1-Persistent CSMA (a=0.01) | ~53% | G ≈ 1.0 |
| 1-Persistent CSMA (a=0.001) | ~75% | G ≈ 1.5 |
1-persistent CSMA significantly outperforms ALOHA, especially when propagation delay is small relative to frame transmission time.
Let's trace through the precise timing of 1-persistent CSMA operation to understand when collisions can and cannot occur.
Scenario: A 2 km cable, propagation speed 200,000 km/s, 1500-byte frames at 10 Mbps.
Case 1: Successful transmission (no collision)
Station A at position 0 km, no other station transmitting or waiting:
| Time | Event | Channel State |
|---|---|---|
| t=0 | A senses idle, starts transmitting | Busy |
| t=10μs | A's signal reaches far end (2 km) | Busy everywhere |
| t=1,200μs | A finishes transmitting | Releasing |
| t=1,210μs | Channel idle everywhere | Idle |
Total time: 1,210 μs. No collision because no other station had data during this window.
Case 2: Propagation delay collision
Station A at position 0 km, Station B at position 2 km, both have data:
| Time | Station A | Station B | Result |
|---|---|---|---|
| t=0 | Senses idle, transmits | Senses idle, transmits | Both start |
| t=5μs | Transmitting | Transmitting | Signals propagating |
| t=10μs | B's signal arrives → COLLISION | A's signal arrives → COLLISION | Both detect |
Collision occurred because both sensed idle within the propagation delay window.
Case 3: Waiting station collision (1-persistent specific)
Station X transmitting. Stations A (at 0 km) and B (at 2 km) waiting:
| Time | Station A | Station B | X | Channel |
|---|---|---|---|---|
| t=0 | Waiting, sensing busy | Waiting, sensing busy | Transmitting | Busy |
| t=1,190μs | X stops at source | Still sensing busy | Done | Transitioning |
| t=1,200μs | Senses idle → Transmits | Still sensing busy | Done | Collision! |
| t=1,205μs | Transmitting | A's signal halfway | Done | Collision! |
| t=1,210μs | B's signal arrives → Collision | Senses idle → Transmits | Done | Collision! |
Even though stations are 2 km apart with 10 μs propagation delay, both transmit within that window because they were both waiting for the channel to clear.
1-persistent CSMA creates a 'synchronization' of waiting stations. They all release simultaneously when the channel clears, maximizing collision probability. This synchronization is what makes 1-persistent suboptimal under high load—the algorithm essentially coordinates stations to collide.
Classic Ethernet (10BASE5, 10BASE2, 10BASE-T in shared hubs) uses a variant of 1-persistent CSMA combined with collision detection (CSMA/CD). Understanding how Ethernet implements 1-persistent behavior provides practical insight.
Ethernet's CSMA/CD with 1-Persistent Carrier Sense:
The IFG as Partial Mitigation:
Ethernet's Inter-Frame Gap provides minimal randomization:
The IFG's primary purpose is receiver recovery and clock synchronization, not collision avoidance. True collision mitigation comes from the exponential backoff after a collision is detected.
Why Ethernet chose 1-Persistent:
Ethernet uses 1-persistent CSMA for INITIAL transmission (sense + immediate transmit), but exponential backoff after collisions. The backoff introduces randomization, breaking the 'waiting station synchronization' for retransmissions. This hybrid works well: simple fast path for typical case, robust recovery for collisions.
Having analyzed 1-persistent CSMA in detail, let's summarize its strengths and weaknesses systematically.
| Scenario | 1-Persistent Suitability | Reason |
|---|---|---|
| Light load LAN | ✅ Excellent | Channel usually idle; immediate transmission optimal |
| Bursty traffic | ✅ Good | Quick channel acquisition; idle periods allow efficient use |
| Moderate steady load | ⚠️ Acceptable | Works but near efficiency limit; consider alternatives |
| Heavy continuous load | ❌ Poor | Guaranteed collisions; throughput collapse |
| Many stations waiting | ❌ Very Poor | All waiting stations collide simultaneously |
1-persistent CSMA represents the most aggressive carrier sensing approach: sense the channel, and if idle, transmit with probability 1 (certainty). If busy, persistently monitor until idle, then transmit immediately.
What's next:
The guaranteed-collision problem of 1-persistent CSMA motivates an alternative approach: Non-persistent CSMA. Instead of camping on a busy channel, non-persistent stations back off immediately and retry after a random delay. This breaks the waiting-station synchronization at the cost of increased idle time.
The next page explores non-persistent CSMA: its algorithm, analysis, and tradeoffs compared to 1-persistent behavior.
You now understand 1-persistent CSMA: its algorithm, collision behavior, throughput characteristics, and real-world implementation in Ethernet. You can analyze when 1-persistent is appropriate (low load, bursty traffic) and when it fails (high load, many waiting stations). Next, we'll explore non-persistent CSMA as an alternative strategy.