An 8086 subroutine that multiplies two signed 16-bit values together without using the mul instruction.

I'm trying to teach myself 8086 assembly.  I need an example of how to use an assembly language subroutine, using only 16-bit registers, to multiply two signed 16-bit values together without using the mul instruction.   My book gives the example below but doesn't give a clear explanation of how to use it:

a) clear the a register
b) place the multiplicand in the D register
c) do the next step 4 times (multiplying 4 bit numbers).  if the left most bit of the D
    register is a 1 then shift both registers to the left one bit, add the multiplier to the
    A register, else shift both registers to the left one bit.