Virginia Tech researchers have developed a broadband channel sounder that can measure channel quality while a wireless network is in operation. Channel measurements from the broadband sounder hold the promise of improving TCP/IP performance by trigging configuration changes in an adaptive data link layer protocol. We present an adaptive data link layer protocol that can use different levels of forward error correction (FEC) codes and link layer automatic retransmission request (ARQ) to improve network and transport layer performance. Using a simulation model developed in OPNET, we determine the effects of different data link layer protocol configurations on TCP/IP throughput and end-to-end delay using a Rayleigh fading channel model. Switching to higher levels of FEC encoding improves TCP/IP throughput for high bit error rates, but increases end-to-end delay of TCP/IP segments. Overall TCP/IP connections with link layer ARQ showed approximately 150 Kbps greater throughput than without ARQ, but lead to the highest end-to-end delay for high bit error rate channels. Based on the simulation results, we propose algorithms to maximize TCP/IP throughput and minimize end-to-end delay using the current bit error rate of the channel. We propose a metric, carrier-to-interference ratio (CIR) that is calculated from data retrieved from the broadband channel sounder. We propose algorithms using the carrier-to-interference ratio to control TCP/IP throughput and end-to-end delay. The thesis also describes a monitor program to use in the broadband wireless system. The monitor program displays data collected from the broadband sounder and controls the settings for the data link layer protocol and broadband sounder while the network is in operation. / Master of Science