Identifying Non-Adjacent Multiuser Allocations By Joint L(1)- Minimization

We consider a device-to-device scenario in a fragmented spectrum band. Multiple devices transmit complex symbols with a single antenna on distributed, but disjoint OFDM resources. The devices select sufficient frequency resources to enable channel estimation, if the receiver has complete knowledge of the resource map. However, in this scenario the actual allocation map is unknown to the receiver. Therefore, a receiver observing the superposition of all transmitted signals is faced with two problems, channel estimation and identification of the correct resources allocation map.In our previous work [1], we identified the channel and the allocation map of the corresponding users by applying an objective function based on the l(1)- norm. In this work, we show that successful recovery is possible under different practical ITU channel models. Furthermore, we apply basis pursuit denoising (BPDN) from the compressed sensing framework for channel estimation and show the superior performance in contrast to classical least square estimators. These results show, that identifying non-adjacent multiuser allocations is possible.