There are more than 17,000 asteroids found near Earth and nearly 2 million asteroids estimated in the main belt between Mars and Jupiter. Asteroid come in diverse forms, some may hold valuable resources such as water, carbon and rare metals that may one day supply a spacefaring civilization. However, asteroids maybe also valuable as relay stations for a permanent high-speed, high-bandwidth interplanetary communication network. Asteroids are typically pock-marked with craters and grooves. Pristine craters resemble a parabolic communication antenna, but without the reflective coating or a receiver/transmitter at the focus. In this work, we evaluate two scenarios, the preliminary feasibility of setting up such a radio antenna on the Martian moon Phobos and Deimos (thought to be captured asteroids) that would act as a communication relay between the Martian system and Earth. Phobos is closer to Mars and is tidally locked. This would require two craters converted to antennas, one perpetually pointing at Mars, another pointing at Earth and a local interconnection between the two. Alternately, the relay on Deimos would need just a single crater relay station. We will then compare this communication relay to the current state-of-the-art, namely the Mars Reconnaissance Orbiter (MRO). The proposed communication antennas would be achieved by landing a swarm of CubeSats onto a crater to form the parabolic reflector. Each CubeSat has a mass of 4 kg and a volume of 3U or 3400 cc with one side forming the surface of the reflector. These CubeSats would hop, roll and fly into the crater and distribute themselves to cover maximum surface area. Each CubeSat has deployable reflectors to fill the gap between adjacent neighbors. A parabolic reflector would be able to reflect radio waves with a gap of one-tenth of the wavelength. A large 12U CubeSat would be positioned at the crater center and extend a deployable tower with a feed antenna to the focus. To achieve the current data rate of MRO, which is 4 Mbps, the power needs of a pair of 20 m2 aperture antennas on Phobos and the interlink will be evaluated. For Deimos, a single 20 m2 antenna will be considered. In both cases, the intent is to have an antenna gain of 50 dBi per crater. The analysis will also be extended to a 200 m2 aperture antenna that can provide a data rate of 40 Mbps and antenna gain of 60 dBi per crater. Our approach to the mission design exploits machine learning to perform formulation, design, planning and operations. The results from these preliminary mission design studies will be used to identify a pathway towards detailed design and field studies in a simulated environment.