Cooperative BS Assignment & Resource Allocation (CBARA)

• CBARA is a joint optimization framework that dynamically assigns base stations and allocates resources to balance communication and sensing in 6G ISAC networks.
• It employs an alternating heuristic optimization algorithm with convex relaxation and thresholding techniques to efficiently allocate transmit power and bandwidth.
• Empirical results show over 100% improvement in communication rates and a 40% gain in sensing accuracy, demonstrating its practical scalability in multi-BS environments.

A Cooperative Base Station Assignment and Resource Allocation (CBARA) strategy for next-generation 6G Integrated Sensing and Communications (ISAC) networks is a joint optimization framework designed to allocate base station (BS) resources—including assignment, transmit power, and bandwidth—across both communication users and sensing targets. The objective is to simultaneously maximize communication throughput and sensing accuracy. The approach in "Cooperative Base Station Assignment and Resource Allocation for 6G ISAC Network" (Liao et al., 12 Sep 2025) formalizes this problem for dynamic, multi-BS, multi-user, and multi-target scenarios, proposes a computationally efficient alternating optimization algorithm, and demonstrates substantial improvements relative to classic non-cooperative schemes.

1. Objective and System Scope

The main objective of the CBARA strategy is the joint optimization of communication and sensing (C&S) performance in a 6G ISAC network comprising multiple BSs, multiple moving targets, and multiple users. Each BS can serve both as a communication and a sensing node, supporting services such as data transmission to users and tracking of moving targets. The CBARA problem poses two primary challenges: dynamically assigning BSs to users/targets (BS assignment), and allocating constrained physical resources (transmit power and bandwidth) in a manner that balances the conflicting objectives of high communication throughput and accurate target localization/tracking.

The network operates under practical resource constraints: each BS and each resource block face transmit power and bandwidth limitations, and only a limited number of BSs may be assigned to each target or user per time slot.

2. Joint Optimization Criteria

Optimization in the CBARA strategy is fundamentally multi-objective and is mathematically characterized by a weighted sum of two metrics:

• Communication performance: Maximized via the aggregate achievable rate for all ISAC users. For user $i$,

$$R_i(p, b) = \sum_{k \in \mathcal{S}_{o,i}} b_{i,k} \log_2\left(1+\frac{p_{i,k} \psi_{i,k}}{b_{i,k} \sigma_c^2}\right)$$

where $p_{i,k}$ and $b_{i,k}$ are power and bandwidth assigned from BS $k$ to user $i$, $\psi_{i,k}$ is a gain factor (array gain, propagation), and $\sigma_c^2$ is the noise variance.

• Sensing performance: Quantified by the predictive posterior Cramér–Rao lower bound (PCRLB) for the position estimate of each target $m$:

$$\mathrm{PCRLB}_{x_m} = \mathrm{tr}\left[J_{x_m}^{-1}(u_m, p_m, b_m)\right]$$

where $J_{x_m}$ is the posterior Fisher information matrix, recursively combining prior motion model information and measurement information from assignments $u_m$ and resource allocations $p_m$, $b_m$.

A global objective is defined as:

$$\min_{U, P, B} \; \eta \sum_{m} \mathrm{tr}\big(J_{x_m}^{-1}(u_m, p_m, b_m)\big) - (1-\eta) \sum_i R_i(u_i, p_i, b_i)$$

where $0 < \eta < 1$ scales the preference for sensing accuracy versus communication throughput.

System constraints enforce maximum power, bandwidth, and the maximum number of assignments per BS and per user/target; all hard (resource) constraints are integrated into the optimization.

3. Algorithmic Solution: Alternating Heuristic Optimization

The resultant CBARA problem is non-convex due to integer (binary) assignment variables, non-linear relationships between sensing/communication metrics, and strong coupling of decision variables. To address this, a two-step heuristic alternating optimization (AO) algorithm is proposed:

Step 1: BS Assignment

• All BS–object assignments are initially considered active.
• A convex resource allocation is performed (with all possible assignments), yielding preliminary power allocations.
• A thresholding mechanism: for each potential assignment, if the preliminary allocated power from BS $k$ to object $m$ exceeds a chosen fraction $\phi$ of BS $k$'s total power, the assignment is activated ($u_{m,k}=1$); otherwise, deactivated ($u_{m,k}=0$).
• This produces a binary assignment matrix $U$ satisfying per-object assignment cardinality constraints.

Step 2: Resource Allocation (Power and Bandwidth)

• With assignments fixed, the remaining problem is to allocate power and bandwidth to each active BS–object link to maximize the (weighted) global objective.
• The trace of the inverse FIM in the objective is relaxed using an auxiliary matrix variable $D_m$ for each target, with a Schur complement constraint:

$$\begin{bmatrix} D_m & I \ I & J(x_m) \end{bmatrix} \succeq 0 \implies D_m \succeq J(x_m)^{-1}$$

• Alternating optimization: iteratively update the power allocation while holding bandwidth fixed, then update bandwidth while holding power fixed, until convergence of the objective.

This alternating process allows efficient search for highly effective solutions without requiring an exhaustive search over all possible assignment/resource configurations. It also allows for adjusting the tradeoff parameter $\eta$ at runtime to prioritize sensing or communication as needed.

4. Mathematical Formulations and Constraints

Fundamental equations forming the core of the CBARA strategy include:

• Achievable Rate:

$$R_i = \sum_{k \in \mathcal{S}_{o,i}} b_{i,k} \log_2\left(1+\frac{p_{i,k} \psi_{i,k}}{b_{i,k} \sigma_c^2}\right)$$

• FIM Recursion for PCRLB:

$$J(x_m^n) = \left(Q_m + F_x J(x_m^{n-1})^{-1} F_x^T\right)^{-1} + \sum_{k \in \mathcal{S}_{o,m}} H_{m,k}^T \Lambda_{m,k}^{-1} H_{m,k}$$

where $F_x$ stems from the object motion model, $Q_m$ and $\Lambda_{m,k}$ are noise covariance matrices, and $H_{m,k}$ is the Jacobian of the measurement from BS $k$.

• Schur Complement for Convexification:

$$\begin{bmatrix} D_m & I \ I & J(x_m) \end{bmatrix} \succeq 0$$

• Global Objective Function:

$$\min_{U,P,B} \; \eta \sum_m \mathrm{tr}(D_m) - (1-\eta) \sum_i R_i$$

subject to power, bandwidth, assignment, and per-link resource constraints.

5. Performance Metrics and Gains

Empirical results demonstrate that the CBARA strategy offers substantial gains:

• A 117% improvement in communication rate compared to baseline uniform allocation.
• A 40% relative improvement in sensing accuracy (as measured by PCRLB reduction).
• The ability to dynamically trade off between these two objectives by varying the scaling parameter $\eta$.
• Simulation studies confirm that the alternating heuristic approach gives performance levels close to those attainable by exhaustive search, but at dramatically lower computational complexity.

These gains are observed across a range of target/user dynamics, heterogeneous BS deployments, and under tight resource constraints.

6. Practical Implications and Extended Applications

CBARA offers a unified resource management approach for future 6G ISAC networks supporting both traditional communication and emerging wireless sensing services. It is directly applicable to:

• Autonomous vehicular networks for joint environment sensing and data exchange.
• Urban surveillance and smart city scenarios involving multi-object tracking and broadband connectivity.
• Multi-cell UAV networks requiring dynamic resource adaptation due to mobility/heterogeneity.
• Any dense multi-BS system requiring both spectral efficiency and precision localization/tracking.

The framework is robust to practical imperfections such as beamforming errors and time/frequency synchronization inaccuracies (explicitly modeled via a degradation coefficient $\delta_t > 1$ in the system equations). Limitations include the iterative computational burden for real-time operation in extremely dense or highly dynamic environments and the reliance on accurate knowledge of noise covariances and gain parameters.

7. Conclusion

The CBARA framework for 6G ISAC networks formulates and solves a joint base station assignment and resource allocation problem that is tailored to the dual requirements of high-data-rate transmission and accurate, real-time sensing. By leveraging alternating optimization, convex relaxation via the Schur complement, and a tunable tradeoff coefficient, it achieves performance improvements (over 100% communication rate and 40% sensing accuracy gains compared to classic benchmarks) while maintaining scalability for practical deployment. CBARA stands as a reference approach for unified 6G network operation, supporting both advanced communication and sensing objectives within a single cooperative multi-BS infrastructure (Liao et al., 12 Sep 2025).