We investigate a $K$ -user parallel packet-erasure broadcast channel. There is an ongoing effort to harness millimeter-wave bands, which are known to be unstable having high outage probabilities, by combining them with stable legacy bands. Motivated by this effort, we consider a heterogeneous scenario in which the parallel subchannels are categorized into two classes having different outage probabilities. For the two-user case, we characterize the sum capacity by developing an explicit achievable scheme and deriving a matching upper bound. In contrast to suboptimal schemes that apply coding on a per-subchannel basis only, our scheme applies coding across subchannels to exploit coding opportunities that arise from asymmetric outage probabilities more efficiently, thereby achieving optimality. By extending our scheme systematically to be applicable for the $K$ -user case, we show that it can provide significant gains over existing schemes. Compared to the $K$ -user scheme currently employed in practice, which allocates chunks of subchannels to users exclusively, we demonstrate the performance improvement attainable by our scheme to be substantial, as the multiplicative gain scales with $K$ . Moreover, we find that our scheme outperforms a per-subchannel extension of state-of-the-art $K$ -user schemes by large margins, further reducing the optimality gap. Our results suggest a potential coding scheme that can be employed in future wireless systems to meet ever-growing mobile data demands.

中文翻译：

---

双类并行分组删除广播信道的总容量

我们调查了 $ K $ 用户并行分组擦除广播信道。通过将毫米波频段与稳定的传统频段相结合，人们正在努力利用毫米波频段，该频段众所周知是不稳定的，具有很高的中断概率。由于这项努力，我们认为异质并行子信道被分为具有不同中断概率的两类的场景。对于两个用户的情况，我们通过开发一个明确的可实现方案并推导匹配的上限来表征总容量。与仅在每个子信道基础上应用编码的次优方案相反，我们的方案应用编码穿过子信道来更有效地利用由非对称中断概率产生的编码机会，从而实现最优性。通过系统地扩展我们的计划以适用于 $ K $ -用户案例，我们证明了它可以比现有方案显着受益。相比 $ K $ -当前在实践中采用的-用户方案，该方案专门为用户分配子信道的大块，我们证明了我们的方案可实现的性能改进是可观的，因为乘法增益与 $ K $ 。此外，我们发现我们的方案优于最新技术的每子通道扩展 $ K $ 用户计划大幅度提高，进一步缩小了最优差距。我们的结果表明，潜在的编码方案可以在未来的无线系统中使用，以满足不断增长的移动数据需求。