Abstract
The two-user multiple-input single-output (MISO) broadcast channel with confidential messages (BCCM) is studied in which the nature of channel state information at the transmitter (CSIT) from each user can alternate between perfect (P), delayed (D), and no CSIT (N). The overall CSIT alternates between nine possible joint states, each occurring for a fraction $\lambda_{I_1 I_2}$ of the total duration. This paper establishes the secure degrees of freedom (s.d.o.f.) region of the MISO BCCM under the symmetry assumption ($\lambda_{I_1 I_2} = \lambda_{I_2 I_1}$). The work highlights synergistic benefits of inter-state coding for secrecy and introduces new converse proofs via local statistical equivalence and channel enhancement.
Introduction and System Model
In wireless communication, the transmitter's knowledge of the channel (CSIT) is a critical resource for managing interference and ensuring message confidentiality. This study focuses on the MISO broadcast channel where a transmitter equipped with two antennas sends private messages $W_1$ and $W_2$ to two single-antenna receivers. A key requirement is secrecy: User 1 must not be able to decode $W_2$, and User 2 must not be able to decode $W_1$.
In practical scenarios, CSIT is rarely static or perfect. It may be delayed due to feedback latency or entirely absent due to channel fading. This paper addresses the complexity of alternating CSIT, where the availability of channel knowledge for each user changes over time or frequency, creating a heterogeneous environment for secure transmission.
Methodology
The research employs a two-pronged approach to characterize the s.d.o.f. region:
Achievability (Inner Bound)
The authors developed a library of "constituent schemes" tailored for specific combinations of CSIT states. By utilizing inter-state coding—coding across different CSIT states rather than treating them in isolation—the transmitter can achieve higher secure rates. This technique leverages the "synergistic gains" where the combination of two different states provides more utility than the sum of their individual parts.
Converse (Outer Bound)
To prove the optimality of these schemes, the authors utilized channel enhancement (providing the transmitter with additional information to create an upper bound) and a novel property termed local statistical equivalence. This property allows virtual receivers to be treated as statistically identical to real ones, simplifying the complex entropy calculations required to bound the secure rate in multi-antenna settings.
Key Results and Insights
The primary contribution is the characterization of the s.d.o.f. region, summarized by the total achievable secure rate. A significant finding is the Marginal Property: for the sum s.d.o.f., only the total fraction of time spent in P, D, or N states (the marginals $\lambda_P, \lambda_D, \lambda_N$) matters, rather than the specific joint states.
The sum s.d.o.f. is established as: $$\text{sum s.d.o.f.} = 2\lambda_P + \lambda_D + \min(\lambda_D, \lambda_N)$$
Key insights include:
- Synergistic Gains: Joint coding across states like Perfect-Delayed (PD) and Delayed-Perfect (DP) can yield a sum s.d.o.f. of 1.5, whereas time-sharing between them would only yield 1.0.
- Saturation Point: If delayed CSIT is abundant ($\lambda_D \ge \lambda_N$), then having no CSIT (N) is as useful as more delayed CSIT for the sum rate. Conversely, if delayed CSIT is scarce, it becomes as valuable as perfect CSIT.
- Cost of Secrecy: The results quantify the rate loss required to ensure confidentiality as a function of the CSIT state distribution.
Conclusion and Future Work
While this work provides the first optimal s.d.o.f. characterization for several heterogeneous CSIT states, it relies on a symmetry assumption between users. Future research is needed to extend these findings to asymmetric CSIT scenarios and to scale the model beyond two users, where the complexity of alternating states increases exponentially. Additionally, investigating non-linear strategies may provide further insights into finite SNR regimes.
Summary updated