We study the fundamental problem of utilization of the channel shared by stations competing to transmit packets on the channel. In their turn, packets are continuously injected into stations' queues by an adversary at a rate at most ρ packets per round. The aim of the distributed medium access control algorithms is to successfully transmit packets and maintain system stability (bounded queues). We further restrain algorithms by introducing an upper bound k on the allowed number of active stations in any given round. We construct adaptive and full sensing protocols with optimal throughput 1 and almost optimal throughput $1-ϵ$ (for any positive ϵ), respectively, in a constant-restrained channel. On the opposite side, we show that restricted protocols based on schedules known in advance suffer from a substantial drop of throughput, at most $\min \{\frac{k}{n},\frac{1}{\log n}\}$. We compare our algorithms experimentally with well-known backoff protocols.

我们研究了在信道上竞争传输数据包的站点共享信道的利用的基本问题。反过来，数据包被对手以每轮最多ρ 个数据包的速率连续注入站的队列中。分布式介质访问控制算法的目的是成功传输数据包并保持系统稳定性（有界队列）。我们通过在任何给定回合中允许的活动站数量上引入上限k来进一步限制算法。我们构建了具有最佳吞吐量 1 和几乎最佳吞吐量的自适应和完整传感协议$1-\epsilon$（对于任何正ϵ），分别在恒定约束通道中。另一方面，我们表明基于预先已知的调度的受限协议最多会遭受吞吐量的大幅下降$\mathrm{分钟}\{\frac{k}{n},\frac{1}{\mathrm{日志}n}\}$。我们通过实验将我们的算法与众所周知的退避协议进行比较。