Gurleen first of all ,it is very difficult to match the results with the paper of others.Because there are some factors which can not mentioned by the author.so getting exact results are very difficult.we can only go to somewhat close to them.As far as i studied the system by varying the freq. Of cpfsk,i found that best results at 1.25 GHz and eye becomes more noisy as we increase it.If you set the freq. More thsn 1.25 GHz i think you will close to results mentioned in paper.But in paper value of ber is very high and q factor is as low as 3 .I will see your implementation and let you about either it is correct or not