Abstract
We study the problem of designing optical receivers to discriminate between multiple coherent states using coherent processing receivers---i.e., one that uses arbitrary coherent feedback control and quantum-noise-limited direct detection---which was shown by Dolinar to achieve the minimum error probability in discriminating any two coherent states. We first derive and reinterpret Dolinar's binary-hypothesis minimum-probability-of-error receiver as the one that optimizes the information efficiency at each time instant, based on recursive Bayesian updates within the receiver. Using this viewpoint, we propose a natural generalization of Dolinar's receiver design to discriminate $M$ coherent states, each of which could now be a codeword, i.e., a sequence of $N$ coherent states, each drawn from a modulation alphabet. We analyze the channel capacity of the pure-loss optical channel with a general coherent-processing receiver in the low-photon number regime and compare it with the capacity achievable with direct detection and the Holevo limit (achieving the latter would require a quantum joint-detection receiver). We show compelling evidence that despite the optimal performance of Dolinar's receiver for the binary coherent-state hypothesis test (either in error probability or mutual information), the asymptotic communication rate achievable by such a coherent-processing receiver is only as good as direct detection. This suggests that in the infinitely long codeword limit, all potential benefits of coherent processing at the receiver can be obtained by designing a good code and direct detection, with no feedback within the receiver.
What Problem Does This Paper Address?
In optical communications, information is typically encoded in "coherent states"—the quantum description of laser light. There are three main ways to catch these signals:
- Direct Detection: Simply counting photons (like a bucket catching rain).
- Coherent Detection: Mixing the signal with a local laser (homodyne/heterodyne) to measure phase.
- Feedback-Enhanced Detection: Using a local laser that "adapts" its intensity and phase based on the photons already received during a single pulse.
The Dolinar receiver is a famous example of the third type; it is the only known receiver that achieves the minimum possible error rate for distinguishing between two coherent states. However, researchers have long wondered if a generalized version of this receiver could reach the Holevo limit—the absolute maximum information rate allowed by quantum mechanics. Reaching this limit is crucial for "photon-starved" applications, such as deep-space communication, where every single photon must carry as much data as possible.
Methods
The authors approach the problem by first re-interpreting the classic Dolinar receiver. Instead of just looking at it as an error-minimizer, they show it is an incremental information optimizer. It uses recursive Bayesian updates to adjust a local oscillator so that every infinitesimal slice of time provides the maximum possible "knowledge" about the incoming signal.
Key methodological steps include:
- Generalization to $M$-ary States: They extend the binary (2-state) logic to $M$ states by using Rényi entropy as a cost function. By tuning the Rényi parameter $\alpha$, the receiver can prioritize "exploring" information early in a pulse and "deciding" on a state late in the pulse.
- Capacity Analysis: They derive the Shannon capacity for a channel using this generalized receiver in the low-photon regime (where the average number of photons per mode, $\mathcal{E}$, is much less than 1).
- Mathematical Proof: They use mathematical induction and entropy bounds to prove an upper limit on the information rate achievable by any receiver using this "semi-classical" feedback architecture.
Key Insights and Ideas
- The "No-Go" Theorem: The most significant insight is a negative result. The authors prove that even the most general coherent feedback receiver cannot outperform standard direct detection in terms of asymptotic capacity.
- The Second-Order Gap: While both direct detection and feedback receivers share the same "first-order" capacity term ($\mathcal{E} \log 1/\mathcal{E}$), they both fall short of the Holevo limit in the "second-order" term. In practical terms, this means a Holevo-achieving receiver could be 20–30 times faster in photon-starved scenarios than the best feedback receiver.
- Information vs. Error: For two states, maximizing mutual information and minimizing error probability lead to the same receiver design. For more than two states, these goals diverge, requiring the use of Rényi entropy to bridge the gap.
- The Necessity of "Truly Quantum" Processing: The results suggest that local feedback is essentially "semi-classical." To beat direct detection, a receiver must use non-classical resources like squeezed light or entanglement within the receiver itself, or perform joint measurements across multiple signal pulses simultaneously.
Key Results
-
Capacity Bound: The achievable rate for a coherent-feedback receiver is bounded by: $$ \frac{C_{coherent}(\mathcal{E})}{\mathcal{E}} \leq \log \frac{1}{\mathcal{E}} - \log \log \frac{1}{\mathcal{E}} + O(1) $$ This matches the scaling of direct detection exactly.
-
Simulation Success: Their generalized receiver (using $\alpha$-entropy optimization) successfully outperformed fixed-parameter receivers in ternary state discrimination, proving their design methodology is sound for short-block hypothesis testing, even if it doesn't boost long-term capacity.
- Peak Power Advantage: Although the capacity is the same as direct detection, the authors note that feedback receivers can work with BPSK modulation, which has much lower peak-power requirements than the "flash-light" OOK (On-Off Keying) required for high-efficiency direct detection.
Limitations and Future Work
- The "Constant Signal" Assumption: The proof assumes the feedback control signal is updated between symbols but remains constant within a single symbol duration. While the authors argue this is a reasonable approximation, a more complex time-varying analysis within the symbol could be explored.
- Non-Coherent Feedback: The study limits the feedback to "coherent" signals (standard laser light). The authors suggest that using squeezed states as local oscillators might finally break the capacity barrier.
- Joint Detection: The paper reinforces that the real frontier is in "Joint Detection Receivers" (JDRs) that process many symbols at once. Designing a practical, realizable JDR remains the "holy grail" of the field.
Summary updated