Overlay Spectrum Access

Unlike underlay or interweave approaches, overlay access requires the secondary user (SU) to possess non-causal knowledge of the primary user's (PU) message, codebook, or transmission parameters. The SU then uses part of its transmit power to relay or assist the primary transmission while superimposing its own data using advanced coding techniques such as Dirty Paper Coding (DPC) or superposition coding. This creates a symbiotic relationship where the PU's effective signal-to-noise ratio improves despite the SU's concurrent transmission, enabling both links to operate simultaneously on the same frequency without mutual degradation.

The theoretical underpinning of overlay access rests on Dirty Paper Coding (DPC), a technique derived from Costa's 1983 information-theoretic result. DPC demonstrates that if a transmitter has perfect, non-causal knowledge of additive interference at the receiver, it can pre-code its signal such that the interference is effectively canceled without consuming additional transmit power. In overlay systems, the SU knows the PU's signal in advance and uses DPC to pre-subtract the primary signal's interference from its own transmission, achieving the same capacity as if the primary user were absent entirely.

Overlay access achieves a strictly larger capacity region than underlay or interweave paradigms. The theoretical cognitive radio channel model demonstrates that both PU and SU can simultaneously achieve non-zero rates—a result impossible in conventional orthogonal sharing. Key capacity characteristics include:

• Rate splitting: The SU divides its message into public and private components
• Cooperative gain: The PU benefits from the SU's relay function, potentially exceeding its standalone capacity
• Pareto-optimal operation: Both users can improve their throughput compared to time-division or frequency-division access schemes

A critical practical constraint of overlay access is the requirement for genuine cognitive capability—the SU must obtain the PU's message or codebook information before transmission. This can be achieved through:

• Backhaul coordination: A wired or out-of-band wireless link between PU and SU transmitters
• Decode-and-forward relaying: The SU first decodes the PU's transmission in one time slot and relays it while superimposing its own data in the next
• Broadcast channel reciprocity: Exploiting the broadcast nature of wireless channels where the SU can overhear PU transmissions intended for its own receiver This requirement fundamentally distinguishes overlay from purely sensing-based interweave approaches.

Practical overlay systems implement cooperation through superposition coding at the SU transmitter and successive interference cancellation (SIC) at the receivers. The SU splits its total transmit power between:

• Cooperative relay power: Used to forward or boost the PU's message, improving the primary link's reliability
• Selfish transmission power: Used to encode the SU's own data using DPC or layered modulation At the PU receiver, the stronger relayed signal is decoded first, and at the SU receiver, SIC strips away the known primary interference before decoding the secondary message. This layered approach enables simultaneous non-orthogonal transmission on the same time-frequency resource.

Modern overlay systems increasingly employ deep reinforcement learning (RL) to optimize the dynamic power-splitting ratio and coding strategy in real time. An RL agent at the SU learns to balance:

• Cooperation level: How much power to allocate to relaying the PU's message based on channel conditions and PU quality-of-service requirements
• Selfish rate maximization: Maximizing its own throughput while maintaining the cooperation constraint
• Channel state adaptation: Adjusting the overlay strategy as fading, mobility, and interference patterns evolve Proximal Policy Optimization (PPO) and Deep Q-Networks (DQN) are commonly used to learn these policies without requiring explicit channel models, making overlay access feasible in dynamic, real-world wireless environments.