Designing systems that simultaneously transmit data and perceive their physical environment—a paradigm known as Integrated Sensing and Communication (ISAC)—introduces a fundamental engineering dilemma. Historically, communication and sensing developed as separate disciplines with entirely different mathematical objectives. Communication systems maximize data throughput, measured in bits per second or Shannon capacity. Sensing systems (like radar) minimize physical parameter estimation errors, measured via Mean Squared Error (MSE), or maximize target detection probabilities, measured via Signal-to-Interference-Plus-Noise Ratio (SINR).
When these two functionalities share the same hardware, spectrum, and waveforms, engineers face an "apples-to-oranges" optimization problem. A waveform optimized purely for communication requires high randomness to carry maximum entropy, whereas a radar waveform often prefers deterministic, structured patterns to minimize target ambiguity. To jointly optimize these systems, researchers have turned to information theory to establish a single, unified mathematical currency: Mutual Information (MI).
Information as a Common Currency
The core intuition behind a unified information-theoretic framework is that both data transmission and environmental sensing are fundamentally tasks of uncertainty reduction.
In a communication channel, the goal is to reduce uncertainty about a transmitted message given the received signal. In a sensing environment, the goal is to reduce uncertainty about a target's state (such as its position or velocity) given the reflected radar echo. By framing both operations as information transfer, we can directly compare and balance them within a single objective function.
As demonstrated by Chongjun Ouyang et al. in Integrated Sensing and Communications: A Mutual Information-Based Framework, adopting mutual information as a single metric bypasses mismatched estimation-theoretic and information-theoretic equations. This allows designers to apply standard information-theoretic optimization tools, such as water-filling algorithms, directly to dual-functional systems.
Traditional Approach:
[Communication] ---> Maximize Data Rate (Bits/Sec) \ No Direct Mathematical
[Sensing] ---> Minimize Estimation Error (MSE) / Bridge for Joint Trade-offs
Unified Information-Theoretic Approach:
[Communication] ---> Maximize Message Mutual Information \ Shared Currency:
[Sensing] ---> Maximize Target Mutual Information / Tractable Pareto Optimization
This shared language makes it possible to map a precise Pareto front between the two functions. Zhiqing Wei et al. leveraged this in Waveform Design for MIMO-OFDM Integrated Sensing and Communication System: An Information Theoretical Approach to derive optimal power allocation strategies that explicitly trade off target information gain against channel data throughput. Similarly, Jinghui Piao et al. extended this framework to uplink systems in Mutual Information Metrics for Uplink MIMO-OFDM Integrated Sensing and Communication System, demonstrating that bitstream recovery and target parameter estimation can be optimized simultaneously under a mathematically consistent framework.
Bridging Information Theory and Physical Accuracy
While maximizing mutual information provides a clean optimization framework, a practical question remains: How does maximizing target mutual information translate to actual physical radar accuracy?
The connection relies on fundamental mathematical links between information theory and estimation theory. As highlighted by Shihang Lu et al. in Integrated Sensing and Communications: Recent Advances and Ten Open Challenges, communication throughput and target estimation accuracy are dual aspects of the same underlying physical process. Specifically, maximizing the mutual information of a sensing target is closely tied to minimizing the Cramér-Rao Bound (CRB), which dictates the lower limit of variance for any unbiased estimator.
To exploit this relationship directly in hardware design, Xinyang Li et al. showed in Optimal Linear Precoder Design for MIMO-OFDM Integrated Sensing and Communications Based on Bayesian Cramér-Rao Bound that the log-determinant of the Bayesian Fisher Information Matrix can serve as a unified metric. This mathematical bridge allows a single optimization objective to bound both parameter estimation errors and communication capacity limits simultaneously.
Furthermore, this framework can accommodate multiple sensing objectives at once. For instance, Ruofeng Yu et al. introduced a "Mutual Information Upper Bound" (MIUB) in Waveform design based on mutual information upper bound for joint detection and estimation to balance the internal radar trade-offs between target detection and parameter estimation without relying on arbitrary, hand-tuned weighting factors. When environmental uncertainty is high, Fengming Xin et al. demonstrated in SINR- and MI-Based Double-Robust Waveform Design that simultaneously maximizing worst-case SINR and MI ensures robust performance even when target and clutter spectra are poorly known.
Algorithmic and Geometric Optimization
Translating these unified metrics into real-time systems requires solving complex, non-convex optimization problems under practical hardware and network constraints.
One major hurdle is computational complexity. Active sensing objectives often rely on metrics like Kullback-Leibler divergence (KLD) to distinguish between hypotheses (e.g., target present vs. absent). To make these calculations scalable, Jeongwoo Park et al. developed a framework in Fractional Programming for Kullback-Leibler Divergence in Hypothesis Testing that uses matrix fractional programming to align sensing objectives with the fractional structures typical of communication metrics like mutual information, drastically reducing optimization runtimes.
Other frameworks incorporate physical and structural constraints directly into the information-theoretic loop:
- Packet and Channel Constraints: Xin Yuan 0004 et al. developed a joint optimization framework in Waveform Optimization for MIMO Joint Communication and Radio Sensing Systems with Imperfect Channel Feedbacks that accounts for real-world packet structures, explicitly modeling training overhead and subsequent channel estimation errors.
- Spatial and Hardware Deployment: In cell-free networks, Yongming Huang et al. combined D-optimality (which minimizes the geometric uncertainty ellipsoid of target localization) with communication SNR metrics in Access Point Deployment for Localizing Accuracy and User Rate in Cell-Free Systems to determine the optimal physical placement of base stations.
- Multi-Base Station Coordination: When scaling to networked environments, Zhiqing Wei et al. demonstrated in Integrated Sensing and Communication enabled Multiple Base Stations Cooperative Sensing Towards 6G how mutual information can manage the interference generated when multiple base stations cooperatively track targets while sharing spectrum with data users.
Ultimately, these optimizations allow researchers to map out the fundamental boundaries of shared resources. By analyzing these systems through the lens of non-orthogonal resource allocation, works such as Revealing the Impact of Beamforming in ISAC by Chongjun Ouyang et al. and the comprehensive review by Yuanwei Liu et al. in The Road to Next-Generation Multiple Access: A 50-Year Tutorial Review chart how beamforming and power multiplexing behave when sensing and communication signals compete for the exact same resource blocks.
The Universality of Uncertainty Reduction
The power of framing performance as the mathematical removal of entropy extends far beyond wireless networks. The underlying principle—that disparate, conflicting indicators can be unified by measuring how much uncertainty they eliminate—is a universal concept.
For example, in biological systems, Vudtiwat Ngampruetikorn et al. demonstrated in Energy consumption and cooperation for optimal sensing that mutual information provides a far more comprehensive assessment of environmental information transfer in cellular sensors than traditional signal-to-noise ratios, capturing complex, energy-consuming cooperative interactions that SNR misses.
Similarly, in medical diagnostics, Shuhan He et al. introduced "entropy removal" in Entropy removal of medical diagnostics as a meta-metric. By applying Shannon entropy to clinical data, they synthesized traditionally conflicting diagnostic indicators like sensitivity and specificity into a single, quantitative value.
Whether evaluating a 6G base station trying to track a vehicle while streaming video, a biological cell sensing a chemical gradient, or a clinician evaluating a diagnostic test, information theory provides the ultimate common denominator: it measures how much closer an observer gets to the truth.
Key Foundational Work
Integrated Sensing and Communications: A Mutual Information-Based Framework (2022) — contributed 2 concept elements to this direction.
Waveform Optimization for MIMO Joint Communication and Radio Sensing Systems with Imperfect Channel Feedbacks (2020) — contributed 1 concept elements to this direction.
Energy consumption and cooperation for optimal sensing (2020) — contributed 1 concept elements to this direction.