The MAC(Media access control) as the name suggests controls PHY when to access the media. All the
scheduling, knowledge of network, beacons, formation of frames, encrypion/decryption, FCS generation/checking is done in the MAC. MAC has knowledge of the whole frame while PHY acts on the bit.
The 802.11 MAC and UWB MAC are similar at a high level. The QoS features are more important in UWB and the higher data rates in UWB will dictate architecture. The protocols also are very different.
From an architecture point of view, the functions need to be split between software and hardware. High-level protocol functions will be implemented in software, low-level functions like encryption/decryption, FCS will be implemented in hardware. And also time critical computations need to be done in hardware.
Power management is also are very important for mobile devices. Several techniques can be used including gated clocks, power gating or back biasing, running on slow clk, using multiple voltages. Power management should be considered at the architecture stage itself to discover potential power savings, For example, running non-critical blocks on slower clock. This brings up an important question. Based on the equation power = k.C.V^2.f.s, where f= frequency and s=switching activity, increasing the frequency decreases the switching activity for a given data path. Does this mean power remains the same when frequency is increased? A good document is http://focus.ti.com/lit/an/spra373a/spra373a.pdf
The answer is yes except that power dissipation in the clk-tree will be higher at higher frequency. See Low Power article on main page for further details.