Source-linked AI summary
An Approach to Asynchronous Unsourced Random Access
Alireza Karami, Dmitry Trukhachev
TL;DR
The paper addresses the lack of prior asynchronous URA work by proposing and evaluating a fully asynchronous transmission system. Its receiver uses timing acquisition with a small preamble pool, and the reported performance is close to that of a slotted URA counterpart.
Problem
The paper addresses the reported lack of prior work on asynchronous URA.
Method
The proposed fully asynchronous URA system combines timing acquisition with payload modulation and a small pool of preambles corresponding to distinct permutation and scrambling choices.
Results
The proposed receiver delivers performance close to that of a slotted URA system under user loads.
Takeaways & Limitations
A preamble pool of 4 outperforms the other two pool choices, while preamble length Np = 30 yields higher overall numbers of supported users.
Abstract
from arXiv · showhide
In this work we propose an approach to construct a fully asynchronous unsourced random access communications (URA) system. Besides the common URA features such as the absence of the user identification information in the packet and the decoding oriented to receiving the content of the messages, there is no limitation on the user transmission timing. The active users can transmit their packet any time on demand. The proposed system belongs to the class of preamble-payload URA formats. Contrary to the typical role of the preamble to serve as a temporary user identification, the preamble serves for the purpose of timing acquisition. The proposed system also employs a very small pool of preambles to resolve collisions of packets transmitted simultaneously.We provide a performance analysis and simulation results for the proposed system and demonstrate that for a wide range of signal-to-noise ratios (SNR)s the supported numbers of active users are close to the numbers of users supported by a slotted URA system counterpart.
I. INTRODUCTION
URA delivers multiple users’ messages to a common receiver without requiring transmitting-user identities, but most existing systems are synchronous and slotted. The paper motivates fully asynchronous access because packets may begin at arbitrary times, making timing acquisition and collision handling central challenges.
- URA targets grant-free communication in which the receiver decodes packet content without requiring transmitting-user identities.
- Most URA systems are synchronous and slotted, restricting packet transmissions to common frame beginnings.
- Asynchronous access can arise from users’ wake-up cycles, on-demand schedules, or pseudo-random transmission delays.
- The paper develops a distinct fully asynchronous URA system because existing NOMA approaches include asynchronous elements but do not yet provide a fully asynchronous solution.
B. Correlation Detection and Machine Learning
Prior work applies machine-learning methods to asynchronous packet-start detection, reporting improvements over correlation-based detection, while also noting that conventional comparisons may be incomplete.
- Machine-learning methods, including neural networks and random forests, have been studied for detecting packet starting times.
- A supervised-learning neural network was reported to achieve 10 times improvement in reducing both false and missed packets for asynchronous short-packet transmission.
- A one-dimensional convolutional neural network has been used as an alternative to correlation-based packet detection in IEEE 802.11.
- The reviewed conventional methods do not receive an appropriate comparison in the cited discussion.
C. Iterative Timing Acquisition and Multiple Access Decoding
The proposed URA system combines asynchronous packet timing with preamble-payload signaling and a small preamble pool used to separate simultaneous arrivals. Its receiver uses timing acquisition and multiple-access decoding, while packet overlaps and collision events are modeled explicitly.
- D. Asynchronous URA: The system removes restrictions on packet starting times, allowing active users to transmit whenever packets are generated.
- System Model: Each packet contains a known preamble for timing acquisition and a payload carrying unsourced-random-access information.
- System Model: Payload modulation uses bit repetition, permutation, and scrambling, while permutations leave the preamble positions unchanged.
- System Model: A small preamble or signature-permutation pool separates packets that arrive simultaneously and would otherwise form collisions.
- System Model: Packets may overlap across adjacent slots, so the received signal includes active packets, additive white Gaussian noise, and packet start indicators.
B. Arrival Times Detection
Packet-start acquisition is required before interference cancellation and error correction. The paper selects a preamble-based detector rather than an energy-based detector for this timing-acquisition step.
- Arrival Times Detection: Acquiring packet starting times is necessary before interference cancellation and error correction can proceed.
- Arrival Times Detection: Two common timing-detection approaches use correlation with a known preamble or energy measurement after permutation and despreading.
- Arrival Times Detection: The proposed system develops the preamble-based approach for detecting packet arrival times.
1) Preamble-based Arrival Time Detection:
The receiver uses preambles to acquire packet arrival times, then recursively combines timing detection with interference cancellation and multi-user decoding. Adaptive thresholds account for changing interference across time segments and iterations.
- Preamble construction and correlation: Preambles are known low-cross-correlation sequences used to detect packet arrival times rather than temporary user identification.The receiver correlates the received sequence with every possible preamble at every symbol shift and declares an arrival when correlation exceeds a threshold.
- Arrival-time detection: The timing acquisition correlator tests each preamble at every shift against a selected correlation threshold.Detection occurs whenever the correlation value for a preamble and shift is larger than Thrcorr.
- Iterative receiver process: Recursive timing detection and interference cancellation are repeated until interference and noise approach the noise power, supporting successful arrival-time detection.Afterward, interference cancellation occupies the processing power for the remaining decoding iterations.
- Threshold setting: Higher thresholds reduce false packets but increase missed packets, whereas lower thresholds detect more real packets while collecting more false packets.Excess false packets can overwhelm the receiver with additional interference.
- Threshold setting: The adaptive threshold assigns lower values to lower-interference segments and higher values to higher-interference segments.Interference increases across the decoding window, while iterative cancellation can reduce the threshold required for a segment.
3) Iterative Timing Detection/MUD with Cyclic Redundancy Check:
The regular iterative timing acquisition and multi-user decoding process can accumulate false-packet interference and degrade or halt receiver operation. ITD/MUD-CRC adds cyclic redundancy checks so detected packets are retained only when they pass validation, and their interference is removed iteratively.
- Motivation: Accumulating false packets increase interference through iterations and can degrade performance or stop the receiver entirely.The receiver therefore modifies the iterative timing acquisition and MUD structure.
- ITD/MUD-CRC: ITD/MUD-CRC combines iterative timing detection and multi-user decoding with a cyclic redundancy check.The CRC is added to the encoded data bits in the modified packet structure.
- ITD/MUD-CRC: CRC bits distinguish real from false packets, and packets passing the check are removed together with their associated interference.Timing acquisition and MUD are repeated for several iterations.
- System depiction: The paper illustrates the iterative timing detection and multi-user decoding process with CRC and contrasts regular and ITD/MUD-CRC packet structures.These roles are summarized in Figures 3 and 4.
4) Collision Reduction:
Asynchronous transmission makes simultaneous packet collisions possible, and collisions can both prevent the collided packets from decoding and add interference for other packets. The system reduces this risk with a small random pool of permutation-signature preamble pairs.
- Collision problem: Without controlled transmission times, simultaneous packet collisions may occur frequently under some parameter settings.A collision occurs when packets share the same permutation, signature, and arrival time.
- Collision effects: Collisions prevent the collided packets from decoding successfully and leave residual interference that degrades other packets.These are the two stated harmful effects of collision.
- Collision reduction: Each transmitting user randomly selects a permutation-signature pair from a pool containing Lp pairs.The pool creates Lp unique preambles.
- Collision reduction: Packets arriving simultaneously can still be decoded successfully when their preambles, and thus their permutation-signature pairs, differ.The random preamble choice provides a chance of separating simultaneous arrivals.
III. APPROACH TO ANALYSIS OF ITERATIVE RECEIVER PROCESSING
The analysis models iterative receiver processing through packet-miss and false-detection probabilities, interference and noise evolution, and collision statistics. It considers regular iterative timing acquisition/MUD and the ITD/MUD-CRC variant while identifying approximation and collision assumptions.
- Detection probabilities: The analysis tracks missed-packet and false-packet probabilities as functions of correlator SNR and iteration-dependent SNR.False detection occurs when no packet is present but correlation exceeds Thrcorr; a miss occurs when a transmitted packet falls below it.
- Interference model: Before MUD begins, interference and noise are modeled as zero-mean Gaussian with power (L −1)Np + σ2Np for L = Ka active users.This approximation is justified using the central limit theorem for large L.
- Iterative interference: The interference model accounts for residual detected-packet interference, undetected-packet interference, false-packet interference, and collision-induced interference across positions and iterations.False-packet powers accumulate from previous and current iterations, while collision terms remain fixed across iterations.
- Analyzed receivers: The analysis evaluates both regular iterative timing acquisition/MUD and the ITD/MUD-CRC approach.The regular approach normalizes each bit power to 1, while the CRC variant uses the proposed iterative receiver structure.
- Model limitation and adjustment: Equation (12) does not appropriately estimate α^(i), so collision-related quantities are adjusted using average collided-packet counts and experimentally computable probabilities.The analysis includes cases with Bp = 10000 and Bp = 30000.
- Collision analysis: Collision statistics estimate concurrent packet arrivals and average collided packets for different active-user loads and preamble bandwidths.The analysis primarily considers two-packet collisions because their probabilities dominate higher-order collision cases.
IV. NUMERICAL RESULTS
The numerical results evaluate asynchronous URA across preamble lengths, preamble-pool sizes, coding choices, false-packet effects, and timing-acquisition conditions. They identify configurations that support more active users while exposing degradation from interference, false packets, and occasional acquisition or cancellation failure.
- Preamble length Np = 30 is reported as a good choice because it produces higher overall numbers of supported users.
- The experiments compare the lowest SNR supporting a given number of concurrently transmitting active users against a controlled-delay system with known packet start times.
- Timing detection and interference cancellation can completely fail under some channel conditions and packet loads, requiring users and the receiver to restart transmission and reception.
- A pool of 4 preambles outperforms pools of 1 and 8 preambles in the evaluated curves.The authors attribute the weaker 8-preamble result potentially to its higher false-alarm probability.
- The coded system performs best at low SNR values, but its longer coded packet reduces the repetition factor M and can degrade higher-SNR performance.The packet-length constraint is approximately 10000 bits in the reported experiment.
- False packets adversely affect performance, especially at higher user numbers corresponding to larger interference.
V. CONCLUSION
The paper proposes and analyzes a fully asynchronous URA system using timing-acquisition preambles, payload coding, collision resolution, and iterative receiver processing. Across a wide range of active-user loads, its performance is close to that of a corresponding slotted system.
- The proposed system supports fully asynchronous URA transmission, with no restriction on when active users transmit packets.It is tested and analyzed for the AWGN channel.
- Timing acquisition uses a preamble, while the payload is modulated through bit repetition, permutation, and scrambling.CRC bits are applied to the payload to facilitate interference cancellation.
- A small preamble pool resolves collisions by associating preambles with distinct payload permutation and scrambling choices.The design uses a small pool rather than assigning a unique preamble to each user.
- The receiver combines iterative timing acquisition and multiple-user detection in a sliding-window architecture.It exploits the natural spatial coupling of the payload graphs.
- Performance remains close to the corresponding slotted URA system across a wide range of active-user loads.
APPENDIX A PROOF OF LEMMA 1
The appendix derives collision probabilities by modeling packet positions as points distributed across lines and counting admissible selections. It then divides the collision-case counts by the total number of selections.
- The proof models packet positions as Lτ lines, each containing nτ points, with points representing possible packet positions.The line structure corresponds to packet positions and possible starting times.
- It first counts selections producing q collisions among the first q lines, using products of the available point counts.The first-instance count is given as N′_1 in equation (25).
- Equivalent counts apply when the second or other packet-position instances from the first q lines are selected, because no instance is preferred.The proof states that N′_2 equals N′_1 and extends the argument to other instances.
- The total number of q-collision cases is obtained by allowing collided points to be selected from different lines.
- The probability of q collisions is computed by dividing the collision-case count by the total number of choices of one point from each of the Lτ lines.