Abstract
Next generation networks are expected to be ultradense and aim to explore spectrum sharing paradigm that allows users to communicate in licensed, shared as well as unlicensed spectrum. Such ultra-dense networks will incur significant signaling load at base stations leading to a negative effect on spectrum and energy efficiency. To minimize signaling overhead, an adhoc approach is being considered for users communicating in the unlicensed and shared spectrums. For such users, decisions need to be completely decentralized as: 1) No communication between users and signaling from the base station is possible which necessitates independent channel selection at each user. A collision occurs when multiple users transmit simultaneously on the same channel, 2) Channel qualities may be heterogeneous, i.e., they are not same across all users, and moreover, are unknown, and 3) The network could be dynamic where users can enter or leave anytime. We develop a multi-armed bandit based distributed algorithm for static networks and extend it for the dynamic networks. The algorithms aim to achieve stable orthogonal allocation (SOC) in finite time and meet the above three constraints with two novel characteristics: 1) Low complexity narrowband radio compared to wideband radio in existing works, and 2) Epoch-less approach for dynamic networks. We establish a convergence of our algorithms to SOC and validate via extensive simulation experiments.
Introduction
As we move toward ultra-dense 5G and 6G networks, the signaling load on base stations is becoming a massive bottleneck. To save energy and spectrum, researchers are looking at "ad-hoc" approaches where devices manage their own connections in unlicensed or shared bands. However, this presents three major hurdles: users cannot coordinate through a central hub, channel quality is heterogeneous (varying by user), and the network is dynamic, with users entering and leaving at will.
Existing solutions often require expensive "wideband" radios that sense all channels simultaneously. This paper introduces decentralized algorithms that allow multiple users to find their own "best" channels using only simple narrowband radios, achieving stable, collision-free allocations even in changing environments.
Methodology: The dSOC Framework
The authors propose two primary algorithms: dSOC_SN (for static networks) and dSOC_DN (for dynamic networks). These algorithms frame channel selection as a multi-player multi-armed bandit (MPMAB) challenge. Coordination is achieved through a structured temporal framework rather than direct communication.
As illustrated in Figure 1, the process begins with a Random Hopping (RH) phase where users jump between channels until they find an empty one to "reserve." This is followed by the Sequential Master and Channel Switching (SMCS) phase. Time is divided into OHS blocks, which are further subdivided into Master Blocks. During these blocks, users take turns acting as the "Master," attempting to switch to better channels ranked by the Upper Confidence Bound (UCB) algorithm.
Key Innovations
Collision as Communication
In a decentralized environment without a control channel, a collision is typically a failure. Here, it is used as a signal. When a Master user attempts to move to an occupied channel, they "collide" with the occupant in a specific pattern. By observing the presence or absence of collisions in dedicated time slots (Channel Transfer and Channel Switch slots), the occupant can decide to accept or deny the swap based on their own preferences.
Epoch-less Dynamics
To handle users entering and leaving, the dSOC_DN algorithm introduces a Synchronization Phase. Instead of restarting the entire network in "epochs" (a common but inefficient approach), a new user "piggybacks" on an occupied channel to observe network timing. By identifying specific silent periods, the new user aligns with the existing block structure without a global clock.
Performance and Stability in Dynamic Networks
The goal of these algorithms is to reach a Stable Orthogonal Configuration (SOC)—a state where no two users want to swap channels because neither would benefit. The paper mathematically proves that these algorithms converge to SOC in finite time.
Figure 3 demonstrates the algorithm's robustness in a dynamic scenario. As users (U1-U4) enter the network or change their sensed channel qualities, the system experiences brief periods of instability (indicated by the shifting lines and sensing dots) before rapidly re-stabilizing. This "epoch-less" approach ensures that the network remains functional and efficient even as the user base fluctuates.
Conclusion
By utilizing narrowband radios and treating collisions as a signaling mechanism, the dSOC algorithms provide a low-complexity solution for decentralized spectrum sharing. Simulations show that these methods achieve higher total rewards and faster stability than previous state-of-the-art wideband approaches, making them ideal for the next generation of ultra-dense, ad-hoc wireless networks.
Summary updated