The Overhead Crisis in Massive IoT
Traditional wireless networks were built for human-centric traffic, such as streaming video or downloading large files. In these scenarios, a device undergoes a complex "handshake" procedure with the base station to establish its unique identity and secure a dedicated channel before transmitting any data.
However, the rise of the Internet of Things (IoT) and Massive Machine-Type Communications (mMTC) has broken the assumptions underlying this design. Imagine a network of millions of environmental sensors. At any given moment, only a tiny fraction of these devices wake up to transmit a sporadic, bursty payload—such as a 100-bit temperature reading. As Alexey Frolov et al. point out in Unsourced Random Access, if a device must execute a traditional handshake protocol to identify itself before sending a 100-bit message, the signaling overhead ends up vastly exceeding the size of the actual data payload.
When millions of uncoordinated devices share a network, assigning unique physical-layer signatures or coordinating access via traditional grant-based protocols becomes an insurmountable bottleneck (Yongpeng Wu et al. in Massive Access for Future Wireless Communication Systems). To scale to millions of devices, networks must abandon the idea of identifying the user before decoding the data.
The Paradigm Shift: Identity-Agnostic Communication
The solution to this bottleneck is Unsourced Random Access (URA), also known as Unsourced Multiple Access (UMAC). The core insight of URA is to completely decouple user identification from data transmission.
In a traditional network, each user has a unique codebook or a unique pilot sequence. In a URA framework, every single device in the network shares the exact same codebook.
As Alejandro Lancho et al. explain in Massive Uncoordinated Access With Random User Activity, all potential users utilize an identical set of codewords to transmit data over a shared channel, such as a Gaussian Multiple Access Channel. Because the codebook is universal, the physical signal carries absolutely no inherent information about who sent it. The communication becomes entirely identity-agnostic.
Why It Works: From Collision Avoidance to Joint Coding
In classical random access protocols (like ALOHA), if two devices transmit simultaneously, it results in a collision, and the data is typically lost. This happens because the receiver tries to resolve the identity of the conflicting transmitters at the physical layer.
As Gianluigi Liva et al. argue in Unsourced Multiple Access: A Coding Paradigm for Massive Random Access, URA shifts the problem from collision avoidance to a joint coding problem.
When multiple uncoordinated devices transmit at the same time, their signals superimpose into a composite signal. Because the base station knows the single master codebook shared by all devices, its decoding task is fundamentally simplified. Instead of searching through millions of user-specific codebooks to see who is talking, the receiver simply searches the master codebook to determine which codewords are active within the received signal space.
Consequently, the output of a URA receiver is not a set of user-attributed data streams, but an unordered list of decoded messages (Mert Özateş et al. in Unsourced Random Access: A Comprehensive Survey).
Traditional Access:
[User 1 Signal] ───> [Check Codebook 1] ───> "User 1 said X"
[User 2 Signal] ───> [Check Codebook 2] ───> "User 2 said Y"
Unsourced Random Access (Shared Codebook):
[Superimposed Signals] ───> [Check Master Codebook] ───> List: {X, Y}
If knowing the specific identity of a device is necessary for the application, that identity is simply treated as a few extra bits inside the data payload itself. By pushing user identification from the physical access protocol into the application layer, the network eliminates the handshake overhead entirely.
Practical Decoding and Energy Efficiency
To make this shared-codebook architecture viable in practice, researchers have developed specialized signal processing and coding techniques.
One prominent approach, introduced by Alexey Frolov et al. in A Polar Code Based TIN-SIC Scheme for the Unsourced Random Access in the Quasi-Static Fading MAC, pairs Polar codes with Successive Interference Cancellation (SIC). By utilizing Expectation-Maximization (EM) clustering, the receiver can perform "on-the-fly" channel estimation. This allows it to decode a codeword, estimate its channel impact, and peel it away from the composite signal, iteratively uncovering the remaining messages in the unordered list without any prior metadata about the transmitters.
Furthermore, because IoT devices are often battery-constrained, energy efficiency is a critical constraint. While complex Gaussian signaling achieves optimal theoretical bounds, Anton Glebov et al. demonstrate in Energy Efficiency of Unsourced Random Access over the Binary-Input Gaussian Channel that practical binary modulation schemes can achieve the exact same fundamental energy efficiency limits for URA. This ensures that low-complexity, low-power hardware can be deployed at scale.
The utility of removing physical-layer identities also extends to multi-functional networks. For example, Mohammad Ahmadi et al. demonstrate in Integrated Sensing and Communications for Unsourced Random Access: Fundamental Limits that a base station can simultaneously recover the unordered list of payloads and determine their spatial origins (localization), providing a highly scalable framework for joint sensing and communication.
Flipping the Script: Coded Downlink Random Access
While URA was originally designed for the uplink (many uncoordinated devices sending data to one base station), the shared codebook principle can also be applied to the downlink (a base station sending data to a small, unpredictable subset of active users).
In a massive network, explicitly addressing which users should listen to a downlink broadcast creates an addressing overhead that scales with the total number of potential users in the system ($n$). To circumvent this, Ryan Song et al. propose a shared codebook strategy for the downlink in Coded Downlink Massive Random Access.
Instead of broadcasting explicit user IDs alongside the data, the base station broadcasts a single index pointing to a long codeword within a shared codebook. Active users are assigned specific positions within that codeword structure. When the base station transmits the codeword index, the active users simply locate their intended messages at their pre-assigned positions.
By leveraging the mathematical symmetries of exchangeable source distributions, Ryan Song et al. proved in Coded Downlink Massive Random Access and a Finite de Finetti Theorem that this shared codebook strategy reduces the addressing overhead to an amount that is completely independent of the total user population $n$.
Whether on the uplink or the downlink, shifting from individual user coordination to shared codebook access provides the scaling laws required to sustain millions of sporadic wireless connections.
Key Foundational Work
Massive Access for Future Wireless Communication Systems (2019) — contributed 2 concept elements to this direction.
A Polar Code Based TIN-SIC Scheme for the Unsourced Random Access in the Quasi-Static Fading MAC (2020) — contributed 1 concept elements to this direction.
Covariance-Based Cooperative Activity Detection for Massive Grant-Free Random Access (2020) — contributed 1 concept elements to this direction.