The clue is in those four arrows forming an X. That's a common way to draw a quadrature hybrid coupler. A signal incident on one input splits its power equally between the two outputs, but with 90 degrees phase difference between them.
If you put the same type of load on both outputs of a quadrature hybrid, then the reflections from those loads will add up in anti-phase to the input port, and arrive in-phase at the other input port. This means that regardless of the loads, as long as they are equal, the input to the hybrid will stay matched and reflection-free. You can see that there is a dummy load on the other hybrid 'input' to absorb these reflections.
Now the stage is set for any type of PIN diode attenuator to be placed in the two positions between the two hybrids. Without the hybrids, you would have to carefully design the resistance of an attenuator formed from series and shunt elements to try to keep it matched. With the hybrids, you can use simple shunt diodes (the A,B,C examples) or more complicated series/shunt diodes (D,E,F examples) as the attenuators, and the hybrids route reflected power into the dummy loads, and not back to the user.
The diagram suggests it's a shunt PIN diode arrangement, and suggesting several are used in parallel. In practice, a common way to use shunt diodes is to place them along a transmission line, spaced λ/4 apart at the highest operating frequency. This spacing does an effective shunt->series conversion on each diode, improving the depth of attenuation that can be obtained, at the higher frequency where the shunt diodes' residual inductance is limiting the maximum attenuation that each diode produces.
