Random Access Procedure in Wireless Networks

• Random Access Procedure (ARP) is a protocol enabling uncoordinated wireless transmissions through a four-step handshake (Msg1–Msg4) to resolve preamble collisions.
• Analytical models reveal that ARP performance hinges on collision probabilities and filtered uplink grant capacity, with metrics like access delay guiding system optimization.
• Advanced ARP enhancements, including algebraic coding, compressed sensing, and ML-based collision detection, improve network reliability in LTE, 5G NR, and IoT deployments.

A random access procedure (ARP) is a foundational protocol in wireless communications, enabling terminals with sporadic or uncoordinated transmission requirements to establish access to a shared channel. ARP is central to contention-based uplink acquisition in LTE, 5G NR, NB-IoT, and emerging paradigms, as well as to advanced random access schemes exploiting physical-layer algebraic properties, compressed sensing, or context-aware contention resolution.

1. Classical ARP Message Structures and Collision Dynamics

In practical systems, ARP typically consists of a four-step contention-based handshake: Msg1 (preamble transmission), Msg2 (random access response), Msg3 (connection request), and Msg4 (contention resolution). The procedure is initiated by a user equipment (UE) transmitting a randomly chosen preamble in a time–frequency slot (PRACH), with key configuration parameters broadcast in system information blocks (e.g., SIB2 in LTE/5G) governing preamble set size, periodicity, window sizes, and retransmission/backoff behavior (Belder et al., 10 Apr 2026).

During Msg1, each UE selects one of $N$ available orthogonal preambles and transmits at the next allowable random access opportunity. The base station detects activated preamble indices without knowledge of the number of contending UEs per index (collisions remain latent at this stage). In Msg2, the base station grants uplink resources to each detected preamble. Collision resolution occurs in Msg3/Msg4: if multiple UEs transmitted the same preamble, their connection requests (Msg3) collide on the same uplink grant, and only singleton successes are ultimately resolved in Msg4. Retransmissions, backoff timers, and power ramping steps regulate subsequent attempts.

The collision probability for a given preamble is: $$P_{\text{coll}} = 1 - \left(1 - \frac{1}{N}\right)^{M-1}$$ where $M$ is the number of simultaneously contending UEs. Expected access delays can be derived as a function of collision probability, window sizes, and retry/backoff logic (Belder et al., 10 Apr 2026, Nielsen et al., 2015).

2. Analytical Modeling of ARP Performance and Bottlenecks

ARP performance, including access delay and outage probability, is driven by two intertwined sources of contention: preamble collisions at the physical layer and grant (or resource) starvation due to limited uplink grant capacity per random access response. Closed-form analyses for LTE ARP in MTC scenarios reveal that the system bottleneck shifts from PRACH contention to RAR–uplink grant capacity ($\mu$ grants per opportunity) at high load (Nielsen et al., 2015, Madueño et al., 2015).

The probability of a one-shot ARP failure, $p_f(\lambda_T)$, decomposes into preamble collision probability $p_c(\lambda_T)$ and uplink-grant failure $p_e(\lambda_A)$: $$p_f(\lambda_T) = 1 - [1 - p_c(\lambda_T)][1 - p_e(\lambda_A)]$$ where $\lambda_T$ is the total attempt rate, $\lambda_A$ is the rate of activated preambles, and $$P_{\text{coll}} = 1 - \left(1 - \frac{1}{N}\right)^{M-1}$$0 is computed via queuing models (e.g., $$P_{\text{coll}} = 1 - \left(1 - \frac{1}{N}\right)^{M-1}$$1 with impatient customers). Retransmission dynamics are captured via fixed-point equations linking new-arrival rate, total ARP attempts, and outage (Nielsen et al., 2015).

Dimensioning analysis shows that increasing random access opportunity (RAO) frequency yields diminishing returns, and raising the uplink grant limit $$P_{\text{coll}} = 1 - \left(1 - \frac{1}{N}\right)^{M-1}$$2 can considerably increase supported MTC load at a given outage target. Overprovisioning RAOs can paradoxically reduce system capacity by causing control-channel congestion (Madueño et al., 2015).

3. Advanced Random Access: Algebraic, Network-Coded, and Compressive Protocols

Beyond classical ARP, novel random-access schemes have been developed to exploit the algebraic structure of collisions or to leverage statistical sparsity in user activity.

Sign–Compute–Resolve ARP: This protocol (Goseling, Stefanović, Popovski) combines three building blocks: physical-layer network coding (PLNC), signature codes, and recursive tree-splitting (Goseling et al., 2016, Goseling et al., 2014). In each slot, active users encode packets with a user signature and payload via a linear code $$P_{\text{coll}} = 1 - \left(1 - \frac{1}{N}\right)^{M-1}$$3:

• On collision, the receiver uses PLNC to recover the sum of all transmitted packets in $$P_{\text{coll}} = 1 - \left(1 - \frac{1}{N}\right)^{M-1}$$4, and, from the sum of signatures, unpacks the colliding user set if the collision size does not exceed $$P_{\text{coll}} = 1 - \left(1 - \frac{1}{N}\right)^{M-1}$$5 (signature code's uniqueness threshold).
• If more than $$P_{\text{coll}} = 1 - \left(1 - \frac{1}{N}\right)^{M-1}$$6 users collide, a tree-splitting mechanism recursively partitions the contenders.
• For $$P_{\text{coll}} = 1 - \left(1 - \frac{1}{N}\right)^{M-1}$$7, throughput approaches 1 user/slot, and the overhead vanishes for moderate-to-large payloads ($$P_{\text{coll}} = 1 - \left(1 - \frac{1}{N}\right)^{M-1}$$8 as $$P_{\text{coll}} = 1 - \left(1 - \frac{1}{N}\right)^{M-1}$$9).

Compressive Random Access: In overload scenarios with sparse activity, a control channel is used to jointly estimate which users are active and their respective channels via compressed sensing recovery (BPDN, CoSaMP) (Wunder et al., 2015). The protocol achieves "one-shot" random access—no handshake or prior synchronization is required—and recovery guarantees are linked to Restricted Isometry Properties of the random control channel measurement matrix. Achievable rates and mis-detection/false-alarm error rates can be explicitly bounded as a function of the sparsity and measurement dimensions.

4. 5G/NR and IoT-Oriented ARP Enhancements

5G, IoT, mMTC, and uRLLC requirements have prompted several ARP enhancements:

Two-step and Early Data Transmission (EDT) ARP: Message reduction by combining preamble and data transmission (Msg1+3, Msg2+4), reducing handshake round-trips and lowering latency (Thota et al., 2019).

Reserved and Dynamic Preamble Allocation: Partitioning the preamble set, either reserving a fraction for high-priority/URLLC users or dynamically adjusting the reserved pool to instantaneous demand, minimizes critical collisions without unduly wasting resources (Thota et al., 2019).

Parallel Preamble Transmission: Dual connectivity allows UEs to concurrently transmit independent preambles to two distinct evolved gNBs, reducing the probability of cumulative collision via spatial diversity (Thota et al., 2019).

Partial Preamble Transmission (PPT): In NB-IoT, splitting preamble time-frequency resources into multiple partial units effectively increases the contention space, reducing collision probability and boosting success probability at high system load (Kim et al., 2017).

Context-Aware and RAPID ARP: RAPID integrates preassigned AS Context IDs and prioritizes two-message resolution for periodic, delay-sensitive IoT traffic. Access Pattern Analyzer (APA) and Markov-model optimization balance collision and resource utilization under mixed ARP schemes (Kim et al., 2022).

ML-based Collision Detection: Supervised learning (notably neural networks post-quantization) can classify singleton vs. collided preambles based on observed power delay profiles, enabling base stations to suppress futile Msg2 grant allocations after collision, thus reducing retransmission overhead and latency under mMTC (Cardenas et al., 29 Oct 2025).

5. Random Access in Non-Terrestrial and Massive MIMO Networks

Non-Terrestrial Networks (NTN) amplify ARP challenges due to long propagation delays, necessitating explicit protocol adaptations (Kodheli et al., 2021):

• Sample–subframe level timing advance at UEs, determined via GNSS.
• Subframe-delayed ARP processing at the base station.
• Extended timer fields in SIBs and new PRACH formats to accommodate large geometric footprints.

For 5G Massive MIMO, random access to pilot sequences is essential to permit channel estimation from a large device pool. Protocols include:

1. Spatial Capture (blind resolution): Discriminates users sharing a pilot via timing or spatial signature extraction.
2. Centralized Collision Resolution: Uses coded pilots with null symbol positions to signal collision states.
3. Distributed Collision Resolution (SUCRe): Devices "self-resolve" based on channel-gain feedback by retransmitting only if judged the strongest at the array.

Massive MIMO enables these approaches due to channel hardening, favorable propagation, and spatial discrimination, greatly improving scalability and reducing average access delays and collision probability (Carvalho et al., 2016).

6. Configuration, Simulation Insights, and Practical Guidelines

Real-life ARP configurations are governed in LTE/5G by system broadcast information (e.g., numRApreambles, prachConfigIndex, ra-ResponseWindowSize). Empirical analysis shows that operators often use uniform ARP configuration across neighboring cells, which increases inter-cell collisions. Simulations varying the PRACH configuration among adjacent cells reduce collision rates by up to 61% and median access delay by up to 42%, highlighting the importance of intelligent, cell-specific ARP parameterization (Belder et al., 10 Apr 2026).

Key configuration guidelines:

• Diversify prachConfigIndex assignments across adjacent cells.
• Adapt preamble formats and timer sizes to deployment environment (dense urban vs. rural).
• Monitor collision and delay metrics via SON/O-RAN controllers to dynamically tune ARP configuration.

7. Practical and Experimental Considerations

Slot-based ARP scheduling in open-source SDR environments, such as OpenAirInterface, can significantly improve random access reliability in large-scale field deployments. Ensuring that msg2/msg3 are transmitted over full DL/UL slots (using a sufficient number of OFDM symbols, typically ≥9 for nLoS or long-range) allows 90–100% end-to-end ARP success even under harsh channel conditions, with specific adjustments to GPIO logic for RF PA/LNA control in time-division duplex (TDD) (Boateng et al., 23 Mar 2025).

This broad spectrum of ARP design, modeling, and enhancements continues to evolve, bridging the needs of massive machine-type communications, ultra-reliable low-latency networking, and highly scalable random access under a variety of network and hardware constraints.