あるいは
#define NUM_BYTES 128
unsigned char a[NUM_BYTES];
...
unsigned char prev4bits = (a[NUM_BYTES - 1] >> 4);
for (int i = 0; i < NUM_BYTES / sizeof(int); i++) {
unsigned char b = (a[(i + 1) * sizeof(int) - 1] >> 4);
((int *)a)[i] <<= 4;
a[i * sizeof(int)] |= prev4bits;
prev4bits = b;
}