Introduction
As we transition toward 6G networks, supporting ultra-massive Machine Type Communications (umMTC) becomes a critical challenge. Traditional grant-based access, where devices request permission before transmitting, introduces excessive latency and overhead for millions of IoT sensors. Grant-free transmission offers a solution by allowing devices to transmit spontaneously, but it risks frequent collisions.
Non-Orthogonal Multiple Access (NOMA) addresses this by enabling the base station to decode overlapping signals using Successive Interference Cancellation (SIC). To implement NOMA without complex coordination, the base station pre-configures specific Signal-to-Noise Ratio (SNR) levels. Users randomly select a level and adjust their transmit power accordingly. This research investigates the optimal design of these SNR levels, comparing "pessimistic" and "optimistic" strategies through the lens of Age of Information (AoI)—a metric for data freshness.
Design Philosophies: Pessimism vs. Optimism
The study evaluates two distinct approaches to setting SNR levels for grant-free NOMA:
- Design I (Pessimistic): This approach assumes a worst-case scenario where all other users might select the most interfering levels. It ensures robustness but requires high transmit power, making it "unaffordable" for many users with poor channel conditions.
- Design II (Optimistic): This approach assumes each SNR level is selected by at most one user. It allows for lower, more accessible SNR levels but is fragile; a single collision can disrupt the entire SIC decoding chain.
The researchers modeled the network as a Markov chain to analyze state transitions and derived closed-form expressions for average AoI, validating these with Monte Carlo simulations.
Key Insights: The SNR Regime Matters
The performance of these designs is highly dependent on the available transmit power. There is a fundamental trade-off between affordability and robustness.
As shown in the figure above, Design II provides significant gains at low transmit power because its lower requirements allow more users to participate. However, as power increases, Design I’s robustness to collisions makes it the superior choice. Interestingly, at low SNR, the "unaffordability" of Design I acts as a natural congestion control, forcing some users to stay silent and reducing collisions.
Performance in Massive Connectivity Scenarios
NOMA consistently outperforms traditional Orthogonal Multiple Access (OMA), particularly as the number of users ($M$) grows.
In scenarios with up to 30 users, the AoI for OMA increases rapidly. In contrast, NOMA-assisted grant-free access maintains much lower AoI. The simulations also highlight that adaptive transmission probabilities can further enhance NOMA's performance, keeping information fresh even in dense deployments.
Future Directions
While this work establishes a clear framework for SNR level design, several avenues for future research remain:
- Hybrid Designs: Developing a design that bridges the gap between optimistic and pessimistic approaches using an auxiliary parameter to minimize AoI.
- CSI Realities: Accounting for Channel State Information (CSI) estimation errors, as the current model assumes perfect local CSI for power scaling.
- Optimization: Determining the optimal transmission probability ($P_{TX}$) for different SNR configurations to maximize network efficiency.
Summary updated