Time for ACK to reach back for a frame $=TT_{frame} + PD + PD + TT_{ack} \\= (2 \times 10^6 \times \frac{1}{2 \times 10^6} + 100 + 100 + 0) ms \\=201 ms$.
Amount of data that could be transferred in $201 ms = 201 \times 10^{-3} \times 2Mbps = 402kb = 201 $packets.
So, for at least 50% utilization the number of frames to be sent $= \left \lfloor \frac{201}{2} \right \rfloor = 101$ which requires 7 bits to represent.
Now, the window size in selective repeat protocol is half the sequence number space (given in any text). So, we need one more bit for sequence number field making it at least 8.