Blog 2026-07-25
Who This Is For: Industrial wireless bridge deployment engineers, WISP network planners, and security surveillance system integrators — anyone who needs to squeeze every bit of efficiency out of wireless medium in PtMP topologies.
Core Issue: Standard 802.11 CSMA/CA delivers less than 30% medium utilization in point-to-multipoint (PtMP) scenarios. Throughput collapses when more than 10 clients are connected, and real-time applications like video surveillance and industrial control can’t get deterministic latency guarantees.
Key Conclusions: iPoll 3’s intelligent polling protocol replaces CSMA/CA contention with active AP-driven scheduling, achieving 350 Mbps TCP throughput with 3-5 ms latency and over 80% medium utilization in an 8-client PtMP setup. Dynamic slot allocation supports 200+ concurrent clients, dual-level frame aggregation boosts throughput 10-15x, and four-level QoS queues deliver deterministic guarantees for mission-critical traffic.
When you scale from a simple point-to-point (PtP) link to a point-to-multipoint (PtMP) network, one fundamental shift happens — multiple clients now share the same wireless medium. The standard 802.11 protocol was originally designed for random-access scenarios, built on the assumption that “the medium is idle most of the time.” But in PtMP networks where multiple clients transmit continuously, CSMA/CA’s contention mechanism becomes the bottleneck.
Three critical weaknesses of CSMA/CA:
In PtMP deployments, standard 802.11 CSMA/CA pushes effective throughput below 30% of the PHY rate when client count exceeds 10. With 10 clients, CSMA/CA medium utilization drops under 30%, collision probability grows exponentially, the hidden station problem causes about 15-25% packet retransmissions, and end-to-end latency jumps from 2 ms at idle to 50-200 ms under load.
For a complete product selection guide, refer to the Industrial Wireless Bridge Complete Guide for full YNW 6 series technical specifications.
The design philosophy behind iPoll 3 boils down to one sentence: Let the AP own the scheduling. This fundamentally eliminates CSMA/CA’s contention overhead.
The key thing to understand about iPoll 3 is that it’s not a “hacked” version of 802.11 — it replaces the MAC-layer medium access control mechanism on top of standard 802.11 PHY. This means iPoll 3 retains all the PHY-layer advantages of 802.11ac (OFDM, MIMO, 256-QAM modulation) while replacing CSMA/CA’s random contention with active polling.
When a new client joins an iPoll 3 network, it first sends a registration request to the AP. The AP assigns a unique Connection ID and records the client’s QoS requirements. Once registered, the client is added to the AP’s polling list. Unlike CSMA/CA, registered clients don’t need to sense the channel before transmitting — the AP handles all channel sensing.
The AP sends poll frames to each registered client sequentially, following the configured scheduling policy. Each poll frame contains the client’s allocated transmission slot. Upon receiving the poll, the client starts transmitting immediately — no CCA (Clear Channel Assessment) or backoff algorithm needed. This design completely eliminates both the hidden station problem and collision retransmissions.
iPoll 3 supports two-level frame aggregation — A-MSDU (up to 7935 bytes) and A-MPDU (up to 64 subframes). The AP can allocate a long enough transmission slot in a single poll cycle, allowing the client to send aggregated multi-frame data in one burst, reducing SIFS (Short Interframe Space) overhead.
iPoll 3’s three-step workflow shifts MAC-layer scheduling from contention-driven to scheduler-driven. A single polling cycle supports up to 256 clients, and frame aggregation boosts per-transmission efficiency by 4-8x (A-MSDU up to 7935 bytes, A-MPDU up to 64 subframes).
Antenna configuration directly affects iPoll 3 protocol performance in practice — see the YNW 6-25 vs YNW 6-N Comparison for details.

The comparison below is based on real-world measurements from YNW 6 series in the 6 GHz band, 80 MHz channel width, 2 km line-of-sight link. CSMA/CA follows the standard 802.11 MAC layer as defined in IEEE 802.11ac-2013 and IEEE 802.11ax-2021; iPoll 3’s active polling scheduling borrows the timeslot concept from TDMA while adding a dynamic adaptive algorithm.
| Dimension | CSMA/CA (Standard 802.11) | TDMA (Standard TDM) | iPoll 3 (Intelligent Polling) |
|---|---|---|---|
| Scheduling Method | Contention-based (random backoff) | Fixed timeslot allocation | Dynamic active polling |
| 8-Client TCP Throughput | ~90 Mbps | ~210 Mbps | ~350 Mbps |
| 16-Client TCP Throughput | ~45 Mbps | ~150 Mbps | ~280 Mbps |
| End-to-End Latency (8 clients) | 50-200 ms | 20-40 ms | 3-5 ms |
| Latency Jitter | High (±30 ms) | Medium (±10 ms) | Low (±2 ms) |
| Medium Utilization | <30% | ~50-60% | >80% |
| Client Capacity | ~30-50 (practical) | ~128 (theoretical) | ~256 (theoretical) |
| Hidden Station Problem | Present | Partially resolved | Fully eliminated |
| QoS Flexibility | Coarse (802.11e/WMM) | Fixed allocation | Fine-grained (per-client/per-flow) |
| Over-the-Air Overhead | High (RTS/CTS/ACK + backoff) | Medium (guard intervals) | Low (poll frame reuse) |
| Dynamic Load Adaptation | Passive (collision triggers backoff) | Poor (fixed slots can’t adjust) | Active (AP adjusts polling order and slot size in real time) |
Key observation: While TDMA beats CSMA/CA in throughput, its fixed-slot allocation is inefficient in asymmetric load scenarios. For example, when one client needs to send heavy traffic while others are mostly idle, TDMA still allocates equal-length slots to idle clients, wasting medium. iPoll 3’s dynamic slot allocation adjusts automatically based on real-time traffic demand — that’s the fundamental reason it averages 30-40% higher throughput than standard TDMA.
Frame aggregation is one of the key technologies behind iPoll 3’s high throughput. The core idea is simple: merge multiple small frames into one large frame before sending, cutting down on header overhead and SIFS waste.
In standard 802.11, every data frame goes through this sequence:
For small packets like VoIP, the payload may be just a few dozen bytes, but the overhead above runs into hundreds of microseconds — medium utilization as low as 5-10%.
| Aggregation Level | Max Size | Efficiency Gain | Best For |
|---|---|---|---|
| No Aggregation (Standard) | 1500 bytes | Baseline | Latency-critical control traffic |
| A-MSDU | 7935 bytes (~5 standard frames) | ~3x | Large file transfer, video streaming |
| A-MPDU | Up to 64 subframes | ~8x | Mixed traffic, small-packet batch |
| A-MSDU + A-MPDU Dual Level | Combined | ~10-15x | Optimal (recommended for daily use) |
Practical tip: On the YNW 6 series management interface, frame aggregation is enabled by default. For mixed-traffic scenarios (video surveillance + data + VoIP), keep dual-level A-MSDU + A-MPDU enabled. For pure real-time traffic like industrial control signals, consider lowering the aggregation level to prioritize latency.
For a complete QoS configuration guide and iPoll 3 tuning parameters, see the Industrial Wireless Bridge Complete Guide.
In PtMP networks, different clients carry wildly different traffic types — one might be streaming video surveillance backhaul (latency and jitter sensitive), while another is just downloading files (latency tolerant). iPoll 3’s QoS mechanism lets the AP dynamically adjust polling order and slot allocation based on traffic priority.
According to the IHS Markit Industrial Wireless Survey 2024, roughly 67% of failures in industrial wireless networks can be traced back to a lack of effective QoS policy. iPoll 3’s four-level queue mechanism is designed specifically to address this pain point.
| Priority | Polling Interval | Typical Traffic | Bandwidth Guarantee |
|---|---|---|---|
| High (Queue 3) | 1-2 ms | Video surveillance, VoIP, industrial control | Guaranteed bandwidth, priority scheduling |
| Medium-High (Queue 2) | 3-5 ms | Enterprise ERP, database sync | Minimum bandwidth guarantee |
| Medium-Low (Queue 1) | 5-10 ms | Web browsing, email | Best effort |
| Low (Queue 0) | 10-20 ms | File download, system updates | Idle bandwidth sharing |
Network Settings → QoS → Enable iPoll 3 QoS
Set minimum guaranteed bandwidth (Mbps) and maximum burst bandwidth for each queue
Map traffic to queues based on any of the following rules:
Use the built-in traffic generator or third-party tools (e.g., iPerf3) to run bandwidth tests
Based on the analysis above, here are the key tuning items when deploying iPoll 3 networks:
Performance verification commands (SSH or Web Terminal):
# View current polling stats
iwpriv ap0 get_polling_stats
# View per-client throughput
iwpriv ap0 get_sta_traffic
# View QoS queue status
iwpriv ap0 get_qos_stats
For the full installation and protocol tuning guide, visit the Industrial Wireless Bridge Complete Guide.

iPoll 3 is a proprietary MAC-layer protocol between YNWMICRO devices. When a YNW AP communicates with a YNW client, it uses active polling mode. However, iPoll 3’s uplink interface is fully compatible with standard 802.11 frame formats — meaning you can connect a YNW bridge’s downstream port to a standard switch or router without any compatibility issues. Non-YNWMICRO devices can’t join an iPoll 3 network directly, but they can communicate through a YNW bridge as standard Ethernet devices.
For more iPoll 3 compatibility information, see the Industrial Wireless Bridge Complete Guide.
The fundamental difference lies in slot allocation. Standard TDMA uses fixed timeslots — every client gets equal transmission time per frame, which is inefficient in asymmetric load scenarios. iPoll 3 dynamically adjusts polling order and slot size based on each client’s real-time traffic demand, so idle clients consume almost no medium time. In measured asymmetric load scenarios (e.g., one client downloading a large file while others are idle), iPoll 3 delivers roughly 40% higher system throughput than TDMA.
iPoll 3 is enabled by default on YNW 6 series with sensible settings, and most deployments won’t need changes. Rare edge cases where you might want to adjust: In a pure PtP topology where the AP serves only one client, CSMA/CA’s contention overhead is practically zero — here iPoll 3’s polling overhead is marginally higher (about 2-3% throughput difference, typically negligible). Also, if your network mixes sub-1 ms-latency industrial control traffic with bulk file transfers, consider dropping the polling interval from the default 5 ms to 2 ms to reduce latency.
Different YNW device models perform differently under iPoll 3 — see the YNW 6-25 vs YNW 6-N Comparison for details.
Data in this article is based on YNWMICRO official technical documentation, Qualcomm QCA 9563/9882 chipset datasheets, and YNWMICRO lab test results. Actual performance may vary depending on environmental conditions, client count, and configuration parameters.