The article introduces a new active queue management scheme called Random Exponential Marking (REM), which aims to achieve high utilization and negligible loss and delay in networks. The key idea of REM is to decouple congestion measures from performance measures such as loss, queue length, or delay. Specifically, REM has two main features: 1. **Match Rate Clear Buffer**: REM attempts to match user rates to network capacity while clearing buffers or stabilizing queues around a small target, regardless of the number of users. 2. **Sum Prices**: The end-to-end marking probability observed by a user depends on the sum of link prices (congestion measures) over all routers in the user's path. The article contrasts REM with Random Early Detection (RED), highlighting how REM achieves these features. REM explicitly controls the update of a congestion measure (price) to stabilize input rates and queue lengths. The marking probability is exponentially increasing in the current price, which is updated based on rate and queue mismatches. This ensures that the end-to-end marking probability is exponentially increasing in the sum of link prices, providing a precise measure of congestion in the path. The authors present simulation results demonstrating that REM can maintain high utilization with negligible loss or queuing delay as the number of users increases. They also discuss how REM can address the issue of TCP performing poorly over wireless links by differentiating between losses due to buffer overflow and those due to wireless effects. The article concludes by emphasizing the potential of REM to improve TCP performance over wireless networks, while noting that further study is needed to understand its transient behavior.The article introduces a new active queue management scheme called Random Exponential Marking (REM), which aims to achieve high utilization and negligible loss and delay in networks. The key idea of REM is to decouple congestion measures from performance measures such as loss, queue length, or delay. Specifically, REM has two main features: 1. **Match Rate Clear Buffer**: REM attempts to match user rates to network capacity while clearing buffers or stabilizing queues around a small target, regardless of the number of users. 2. **Sum Prices**: The end-to-end marking probability observed by a user depends on the sum of link prices (congestion measures) over all routers in the user's path. The article contrasts REM with Random Early Detection (RED), highlighting how REM achieves these features. REM explicitly controls the update of a congestion measure (price) to stabilize input rates and queue lengths. The marking probability is exponentially increasing in the current price, which is updated based on rate and queue mismatches. This ensures that the end-to-end marking probability is exponentially increasing in the sum of link prices, providing a precise measure of congestion in the path. The authors present simulation results demonstrating that REM can maintain high utilization with negligible loss or queuing delay as the number of users increases. They also discuss how REM can address the issue of TCP performing poorly over wireless links by differentiating between losses due to buffer overflow and those due to wireless effects. The article concludes by emphasizing the potential of REM to improve TCP performance over wireless networks, while noting that further study is needed to understand its transient behavior.