Non-Primary Channel Access (NPCA)

Updated 4 July 2026

NPCA is a resource-allocation concept where primary users have default rights while secondary users access additional channels under specific conditions.
It is applied in femtocells, cognitive radio, Wi-Fi (IEEE 802.11bn), and 5G NR to improve spectrum efficiency and manage interference.
Design trade-offs in NPCA involve balancing reserved and shared channels, handling switching overhead, and optimizing dynamic access protocols.

to=arxiv_search.search อาคารจีเอ็มเอ็มարկdown both {"query":"\"Non-Primary Channel Access\" IEEE 802.11bn arXiv", "max_results": 10, "sort_by": "submittedDate"} to=arxiv_search.search 北京赛车能_markdown બંને {"query":"NPCA Wi-Fi 8 802.11bn arXiv", "max_results": 10, "sort_by": "submittedDate"} to=arxiv_search.search 彩票天天乐իչ_markdown બંને {"query":"(Zhong et al., 2011) femtocell hybrid access stochastic geometry", "max_results": 5, "sort_by": "relevance"} to=arxiv_search.search 天天中彩票人工 markdown {"query":"(Wei et al., 2024)", "max_results": 5, "sort_by": "relevance"} Non-Primary Channel Access (NPCA) can be understood as a family of access mechanisms in which transmission remains anchored to a primary channel, cell, node, or licensed resource, while additional or alternative resources are made available under explicit restrictions such as hybrid sharing, opportunistic sensing, dynamic channel bonding, or multi-connectivity. In the literature, this idea appears as multi-channel hybrid access in femtocells, opportunistic interweave access in cognitive radio, dynamic use of secondary subchannels in IEEE 802.11ax/11bn, and access to non-primary serving cells or nodes in 5G New Radio (Zhong et al., 2011, Jiang et al., 2011, Wei et al., 2024, Mahmood et al., 2019).

1. Conceptual scope and terminology

The term NPCA is not used uniformly across the cited literature, but the structural pattern is consistent. A primary entity has guaranteed or default access rights to a resource, while a secondary, non-primary, or nonsubscriber entity obtains access only on a subset of resources, only under sensed opportunity, or only when scheduled by a primary anchor. In femtocells, subscribers are the primary users and inside nonsubscribers are the non-primary users. In cognitive radio, the primary user owns the licensed channel and the secondary user opportunistically accesses it. In Wi-Fi, the primary object is the 20 MHz primary channel, while non-primary access refers to secondary subchannels within a bonded channel. In NR, the primary object is the control or PDCP anchor, and non-primary access is realized through secondary nodes, SCells, PSCells, CoMP points, or a decoupled uplink node (Zhong et al., 2011, Jiang et al., 2011, Wei et al., 2024, Mahmood et al., 2019).

Context	Primary resource	Non-primary access form
Femtocells	Reserved subchannels $M_r$ for subscribers	Shared subchannels $M_s$ for inside nonsubscribers
Cognitive radio	PU licensed channel	SU interweave access during OFF periods
Wi-Fi	Primary 20 MHz channel	Secondary/non-primary subchannels via DBCA/NPCA
5G NR	PCell, MN, PDCP anchor	SCells, PSCell, SNs, CoMP points, decoupled UL

This cross-domain mapping matters because it prevents an overly narrow interpretation of NPCA as a Wi-Fi-only feature. A plausible implication is that NPCA is better viewed as a resource-allocation principle: primary anchoring with conditional access to additional resources.

2. Canonical mechanisms in femtocells and cognitive radio

In two-tier femtocell networks, NPCA appears as multi-channel hybrid access. The core definition is explicit: an FAP allows covered nonsubscribers to connect via a subset of available subchannels and reserves the remaining subchannels for subscribers. If an FAP has $M$ orthogonal downlink subchannels, then $M_s$ are shared and $M_r=M-M_s$ are reserved. Each FAP independently and randomly chooses which $M_s$ subchannels are shared. A subscriber selects one reserved subchannel uniformly at random, while an inside nonsubscriber selects one shared subchannel uniformly at random. Closed access corresponds to $M_s=0$ , open access corresponds effectively to $M_s=M$ , and hybrid access corresponds to $0<M_s<M$ (Zhong et al., 2011).

The femtocell analysis is explicitly spatial. Macrocell base stations form a homogeneous PPP $\Phi_m$ of density $M_s$ 0, while femtocell access points are modeled either as a homogeneous PPP $M_s$ 1 of density $M_s$ 2 or as a Neyman–Scott cluster process. Subscribers and inside nonsubscribers are PPPs inside each femtocell disk. This construction allows NPCA to be expressed through the fraction $M_s$ 3, which simultaneously controls admission of inside nonsubscribers, time-sharing pressure on subscribers, and interference through the number of busy shared subchannels (Zhong et al., 2011).

In cognitive radio, NPCA takes the form of opportunistic interweave access under unknown primary behavior. The primary channel is modeled as an ON–OFF process with exponentially distributed $M_s$ 4 and $M_s$ 5, and the secondary network is coordinated through a controller that senses perfectly, maintains a FIFO request list, and authorizes fixed-duration transmissions of length $M_s$ 6 only while the channel is OFF. Because the secondary users are half-duplex and cannot be interrupted once transmission begins, interference occurs if the primary returns during an ongoing secondary transmission. The paper proves that the secondary communication behavior is a renewal process and derives closed-form interference expressions from renewal theory (Jiang et al., 2011).

A central clarification follows from that model: perfect sensing does not imply zero interference. The interference is caused by asynchronous access and the inability to terminate a fixed-length transmission mid-course when the primary reappears. This directly distinguishes NPCA under unknown primary protocol from slotted interweave models in which interference is attributed mainly to sensing error (Jiang et al., 2011).

3. WLAN NPCA from DBCA to IEEE 802.11bn

In IEEE 802.11ax-style dynamic bandwidth channel access, a station contends on its primary 20 MHz channel and then opportunistically extends transmission over additional 20 MHz secondary channels if they are sensed idle. This is functionally equivalent to NPCA on secondary channels. The paper on IBAC shows that this behavior creates a specific collision mechanism, the Outside Warning Range Problem (OWRP): RTS/CTS is exchanged on the primary 20 MHz, but after channel bonding the interference range increases with bandwidth while the transmission range shrinks. A transmitter that did not hear the warning exchange may therefore interfere with the wider bonded transmission. IBAC addresses this by adapting the bonding level through a Markov-chain throughput model and a Thompson-sampling-based Bayesian selector, rather than always choosing the widest locally idle bandwidth (Karmakar et al., 2022).

The 802.11bn literature makes NPCA explicit. One analytical model considers a two-channel abstraction with a primary channel of occupancy $M_s$ 7 and a non-primary channel of occupancy $M_s$ 8. In the legacy model, throughput is

$M_s$ 9

whereas idealized overhead-free NPCA yields

$M$ 0

That result formalizes the intuition that NPCA is valuable when the primary is often blocked and a secondary remains available. The same work then introduces a switching-overhead factor $M$ 1 and shows that switching overhead can be large and can negate the throughput gain, explicitly challenging the assumption that NPCA is invariably superior to legacy access (Wei et al., 2024).

A second 802.11 UHR analysis studies NPCA without modeling switching overhead and derives a multi-channel throughput expression in which the NPCA term is strictly positive under the stated assumptions. In simulation, that model reports that the NPCA network outperforms the legacy network by increasing at least 50% average throughput while reducing at least 40% average delay (Wei et al., 2024).

Recent 802.11bn work also moves from static analytical abstractions to protocol-level and system-level models. One CTMC analysis of dense OBSS deployments shows that NPCA can significantly improve throughput and reduce access delays in favorable conditions and can mitigate the OBSS performance anomaly, but also that it may increase contention on secondary channels and reduce transmission opportunities for BSSs operating there (Bellalta et al., 22 Apr 2025). A simulator-oriented implementation describes NPCA as a mechanism that allows an AP to temporarily translate its operation to a different non-primary channel within its configured BSS bandwidth, triggered by inter-BSS transmissions on the primary channel; the AP then runs a new backoff on the NPCA primary, uses an ICF/ICR exchange before transmission, and is forced to return to the original primary after the NPCA session (Wilhelmi et al., 24 Jun 2026).

These studies establish that Wi-Fi NPCA is not merely “wider is better.” It is a primary-centric MAC extension with explicit trigger conditions, secondary-channel contention, non-negligible protocol overhead, and topology-dependent interference consequences.

4. Multi-channel access in 5G New Radio as NPCA

The NR literature does not generally use the NPCA label, but the mapping is direct. In multi-connectivity and dual connectivity, the master node or PDCP-anchor node is the primary anchor, while secondary nodes provide non-primary channels used either for throughput-oriented packet splitting or reliability-oriented duplication. In the reliability-oriented mode, a packet arriving at the PDCP anchor is duplicated and forwarded over Xn to one or more secondary nodes; the UE retains the first successfully received copy and discards duplicates. The proposed duplication status report allows the UE to inform all nodes in the duplication set once one copy is successfully received, so that pending transmissions on other links can be discarded (Mahmood et al., 2019).

Carrier aggregation provides a second NR realization of NPCA. The PCell is the primary channel, while PSCell and SCells are non-primary channels aggregated at MAC. The paper’s rule-based component-carrier selection mechanism uses RSRQ and CC load, and numerical evaluation reports a median throughput gain of up to 100%, with about 66% gain at the 5th percentile and up to 75% gain at the 95th percentile. At two component carriers, roughly half of UEs use both carriers from the same node, while the other half use carriers from different nodes, making CA and DC simultaneously relevant (Mahmood et al., 2019).

Downlink–uplink decoupling provides an uplink-oriented version of NPCA. The downlink cell remains the primary anchor, but the uplink can be served by a different node chosen from a metric combining radio pathloss and MEC computational proximity. Under the reported setting with inter-tier resource disparity parameter $M$ 2, the MEC-aware DUDe scheme achieves about 40% reduction in median extended packet delay budget (Mahmood et al., 2019).

Coordinated multi-point extends NPCA into the spatial domain. Additional transmission points function as non-primary resources, either through joint transmission for diversity or through coordinated scheduling and interference cancellation. In the reported two-gNB low-latency scenario, the combined MC/CoMP design reduces average two-way latency by 60% relative to the single-connectivity baseline (Mahmood et al., 2019).

Taken together, these NR mechanisms broaden NPCA beyond channel occupancy alone. The non-primary resource may be an additional carrier, a secondary node, an extra transmission point, or a decoupled uplink endpoint, but the organizational principle remains primary anchoring plus conditional exploitation of additional resources.

5. Analytical trade-offs, objective functions, and design knobs

Across the literature, NPCA is quantified through three coupled quantities: resource occupancy, interference exposure, and service sharing. In hybrid-access femtocells, the average number of busy subchannels per FAP is

$M$ 3

which induces the busy-subchannel probability

$M$ 4

and hence the thinned interfering intensity $M$ 5. Increasing $M$ 6 improves access for inside nonsubscribers, but it also increases busy-channel probability and therefore co-tier and cross-tier interference (Zhong et al., 2011).

The same paper makes the subscriber versus nonsubscriber trade-off explicit. Increasing $M$ 7 improves the inside-nonsubscriber rate $M$ 8, reduces the reserved pool $M$ 9, and can worsen subscriber rate $M_s$ 0. Under the default parameters with $M_s$ 1, a stable compromise region occurs for roughly $M_s$ 2, where both overall nonsubscriber rate and subscriber rate are close to their respective maxima (Zhong et al., 2011).

In cognitive radio NPCA, the key interference metric is the fraction of PU ON-time overlapped by SU transmissions,

$M_s$ 3

For non-saturated arrivals, the paper shows that

$M_s$ 4

where $M_s$ 5 is the busy fraction of the secondary queueing system. The design variables are the SU transmission time $M_s$ 6 and the mean inter-arrival time $M_s$ 7, optimized to maximize SU average rate subject to a PU QoS constraint $M_s$ 8 and the queue-stability constraint $M_s$ 9. The reported structural result is monotonic: longer SU packets and higher activity increase SU throughput, but reduce PU rate and tighten the stability constraint (Jiang et al., 2011).

In Wi-Fi 8, the main design knob is no longer just whether NPCA is enabled, but when it is worth paying the switching cost. The optimized 802.11bn study reports average throughputs of 9.2575 Mbps for legacy, 9.5250 Mbps for always-on NPCA, and 10.7251 Mbps for a hybrid occupancy-aware controller in a time-varying two-channel scenario with $M_r=M-M_s$ 0. This makes the point sharply: the relevant comparison is no longer legacy versus ideal NPCA, but static versus adaptive NPCA (Wei et al., 2024).

A broader conclusion follows. NPCA is not a single monotone mechanism. It is a tunable operating point in which better offloading, additional bandwidth, or higher diversity must be balanced against reduced primary-only protection, more secondary-channel contention, and sometimes substantial protocol overhead.

6. Assumptions, misconceptions, and adaptive control

Several recurrent misconceptions are addressed directly by the cited work. One is that NPCA is invariably superior to primary-only access. The 802.11bn overhead analysis rejects that view: switching overhead can be large and can erase the theoretical advantage of non-primary transmission, which is why the authors develop a hybrid model rather than recommending always-on NPCA (Wei et al., 2024). A second is that perfect sensing eliminates interference in cognitive-radio NPCA. In the renewal-theoretic model, perfect sensing is assumed precisely to isolate a different effect: interference persists because fixed-length half-duplex SU transmissions cannot be interrupted when the primary reappears (Jiang et al., 2011). A third is that, in femtocell hybrid access, subscribers and inside nonsubscribers necessarily experience different SINR distributions after channel allocation. The stochastic-geometry model states the opposite: once a femtocell user is allocated a subchannel, the propagation and interference environment is the same; the performance difference arises from channel subset eligibility and time sharing rather than from a distinct SINR law (Zhong et al., 2011).

The assumptions behind current NPCA analyses are strong. They include full-buffer or saturated traffic, orthogonal subchannels, Rayleigh fading, exponential ON–OFF primary activity, perfect sensing, no hidden terminals, single-PU single-channel abstractions, constant transmit powers, static channel partitions, and downlink-only or single-server viewpoints, depending on the domain (Zhong et al., 2011, Jiang et al., 2011, Wei et al., 2024). This suggests caution when extrapolating numerical thresholds across technologies.

Recent WLAN work also indicates how NPCA may evolve from static design to online control. A multi-armed bandit formulation treats each action as the triple

$M_r=M-M_s$ 1

where $M_r=M-M_s$ 2 selects the operational channel $M_r=M-M_s$ 3, $M_r=M-M_s$ 4 selects the primary channel $M_r=M-M_s$ 5, and $M_r=M-M_s$ 6 selects the contention window. In that framework, contextual MABs consistently outperform non-contextual ones, cooperative multi-agent architectures converge faster than single-agent ones, and decentralized learners can exhibit greediness and policy-chasing dynamics that degrade coexisting networks’ performance (Casasnovas et al., 13 Nov 2025). A plausible implication is that future NPCA controllers will be judged not only by throughput gain, but also by fairness, coexistence stability, and the quality of the context used to activate or suppress non-primary access.

NPCA therefore occupies an intermediate position between strict primary protection and unrestricted spectrum use. Its importance lies not in any single instantiation, but in the recurring architectural idea that primary anchoring need not imply primary exclusivity, provided that the resulting access, interference, and control loops are modeled with enough fidelity to expose where the additional degrees of freedom are genuinely beneficial.