This paper investigates the capacity of wireless ad hoc networks, focusing on the interaction between ad hoc forwarding and the 802.11 MAC protocol. The authors use simulations and first-principles analysis to examine how network size, traffic patterns, and local radio interactions affect achievable capacity. Key findings include: 1. **802.11 MAC and Ad Hoc Forwarding**: The 802.11 MAC protocol, which includes RTS/CTS/Data/ACK exchanges, is shown to achieve a throughput of about half the theoretical maximum in a single-cell network. In more complex scenarios, such as chains and lattices, the 802.11 MAC often fails to optimize transmission schedules, leading to reduced capacity. 2. **Traffic Patterns and Scalability**: The paper explores how different traffic patterns impact per-node capacity. Local traffic patterns, where nodes communicate primarily with nearby nodes, allow for constant per-node capacity as the network grows. In contrast, non-local traffic patterns, where the average distance between source and destination nodes increases with network size, result in a rapid decrease in per-node capacity. 3. **Scaling Relationships**: The authors derive scaling relationships for per-node capacity, showing that it is proportional to the square root of the network size for random traffic patterns. They also discuss how power-law distance distributions affect capacity scaling, with patterns that decay more rapidly than a certain threshold maintaining constant per-node capacity. 4. **Related Work**: The paper reviews previous studies on ad hoc network capacity, including work by Gupta and Kumar, Shepard, and Grossglauser and Tse, highlighting the differences in their assumptions and findings. 5. **Conclusion**: The key factor determining the feasibility of large ad hoc networks is the locality of traffic. The paper provides criteria to distinguish traffic patterns that allow scalable capacity from those that do not, offering insights for both simulation studies and real-world network deployment.This paper investigates the capacity of wireless ad hoc networks, focusing on the interaction between ad hoc forwarding and the 802.11 MAC protocol. The authors use simulations and first-principles analysis to examine how network size, traffic patterns, and local radio interactions affect achievable capacity. Key findings include: 1. **802.11 MAC and Ad Hoc Forwarding**: The 802.11 MAC protocol, which includes RTS/CTS/Data/ACK exchanges, is shown to achieve a throughput of about half the theoretical maximum in a single-cell network. In more complex scenarios, such as chains and lattices, the 802.11 MAC often fails to optimize transmission schedules, leading to reduced capacity. 2. **Traffic Patterns and Scalability**: The paper explores how different traffic patterns impact per-node capacity. Local traffic patterns, where nodes communicate primarily with nearby nodes, allow for constant per-node capacity as the network grows. In contrast, non-local traffic patterns, where the average distance between source and destination nodes increases with network size, result in a rapid decrease in per-node capacity. 3. **Scaling Relationships**: The authors derive scaling relationships for per-node capacity, showing that it is proportional to the square root of the network size for random traffic patterns. They also discuss how power-law distance distributions affect capacity scaling, with patterns that decay more rapidly than a certain threshold maintaining constant per-node capacity. 4. **Related Work**: The paper reviews previous studies on ad hoc network capacity, including work by Gupta and Kumar, Shepard, and Grossglauser and Tse, highlighting the differences in their assumptions and findings. 5. **Conclusion**: The key factor determining the feasibility of large ad hoc networks is the locality of traffic. The paper provides criteria to distinguish traffic patterns that allow scalable capacity from those that do not, offering insights for both simulation studies and real-world network deployment.