Human body communication (HBC), which uses the human body tissue as transmission medium to transfer health informatics, serves as a promising physical layer solution for the wireless body area network (WBAN). The channel characteristics and communication performance of galvanic coupling HBC are investigated in this thesis. Firstly, the empirical measurements on the human limb HBC channels from both static and dynamic limb behavior are conducted. Thereafter, a transfer-function channel model is developed. Based on the model, the channel capacity is derived and solved by water-lling algorithm. Moreover, the bit error rate (BER) performance of modulation schemes is derived and veried in experiments. The results show that the human limb channel in short transmission distance (i.e.,<20 cm) behaves as bandpass prole. The channel gain is around -25 dB at 5 cm, an increase of distance 5 cm leads to additional channel attenuation around 10 dB, which causes the channel capacity decrease 50-90% with low transmission power. With distance longer than 20 cm, channel gain in passband is around -55 dB, and channel capacity is several ten kbps with transmission power -20 dBm. The exion of elbow joint causes channel gain increase around 3-5 dB, which causes BER variate around 1- 5 orders of magnitude. While the other eects, such as hand loading conditions and muscle fatigue have negligible eect (p >0.793) on channel gain. The BER in galvanic coupling HBC channel is found to be the same as that in the additive white Gaussian noise (AWGN) channel. Finally, the results about communication performance have been applied in applications to guide the selections of transmission power, data rates and modulation methods. Beyond this, the channel characteristics in dynamic limb behavior are applied to develop a new method for elbow-joint-angle (EJA) estimation. Compared to the myoelectric signal based methods, the proposed idea reveals its advantages in complexity, accuracy and free from inuence of muscle fatigue.