This work formulates a collaborative decision-making framework that exploits over-the-air computation to efficiently aggregate soft information from distributed sensors. This new AirCompFDM protocol approximates the sufficient statistic (SS) of optimum binary hypothesis testing at a server node in this distributed sensing environment under different operation constraints. Leveraging pre/post-processing functions on over-the-air aggregation of sensor log-likelihood ratios, Air-CompFDM significantly improves bandwidth efficiency with little detection loss, even from modest numbers of participating sensors and imperfect phase pre-compensation. Without phase pre-compensation, the benefit of over-the-air sensor aggregation diminishes but still can mitigate the effect of channel noise. Importantly, AirCompFDM outperforms the traditional bandwidth-hungry polling scheme, even under low SNR. Furthermore, we analyze the Chernoff information and obtain the approximate effect of sensor aggregation on the probability of detection error that can help develop advanced detection strategies.

Multiuser MIMO (MU-MIMO) technologies can help provide rapidly growing needs for high data rates in modern wireless networks. Co-channel interference (CCI) among users in the same resource-sharing group (RSG) presents a serious user scheduling challenge to achieve high overall MU-MIMO capacity. Since CCI is closely related to correlation among spatial user channels, it would be natural to schedule co-channel user groups with low inter-user channel correlation. Yet, establishing RSGs with low co-channel correlations for large user populations is an NP-hard problem. More practically, user scheduling for wideband channels exhibiting distinct channel characteristics in each frequency band remains an open question. In this work, we proposed a novel wideband user grouping and scheduling algorithm named SC-MS. The proposed SC-MS algorithm first leverages spectral clustering to obtain a preliminary set of user groups. Next, we apply a post-processing step to identify user cliques from the preliminary groups to further mitigate CCI. Our last step groups users into RSGs for scheduling such that the sum of user clique sizes across the multiple frequency bands is maximized. Simulation results demonstrate network performance gain over benchmark methods in terms of sum rate and fairness.

The tremendous growth of mobile networking and Internet of Things (IoT) demands efficient and reliable service for massive wireless systems. Multi-input-multi-output (MIMO) technologies successfully utilize spatial diversity to substantially improve spectral efficiency by scheduling multiple devices for simultaneous spectrum access. Efficient solutions to the NP-hard problem of scheduling large number of users are vital to interference mitigation and spectrum efficiency. Despite successes of machine learning in tackling large-scale optimization problems, direct adoption of supervised learning in MIMO user scheduling is difficult as there is no optimum solution to use as labeled training data, and unsupervised learning would identify similar user channel features instead of promoting channel diversity. In this work, we propose an effective and scalable user scheduling paradigm based on unsupervised learning to enhance spatial diversity in both uplink and downlink. Given users’ channel state information (CSI), we first cluster CSIs over the Grassmannian manifold to identify users with high CSI similarity, before scheduling them into MIMO access groups with low co- channel interference. Our paradigm is generalizable to a variety of different simple and scalable unsupervised learning tools and different diversity optimization criteria. Numerical tests demonstrate substantial gain in terms of spectrum efficiency and interference suppression at modest computation complexity.

We propose an over-the-air learning framework for collaborative decision making in wireless sensor networks. The low complexity framework leverages low-latency sensor transmission for a decision server to coordinate measurement sensors for hypothesis testing through over-the-air aggregation of sensor data over a multiple-access channel. We formulate several collaborative over-the-air hypothesis testing problems under different practical protocols for collaborative learning and decision making. We develop hypothesis tests for these network protocols and deployment scenarios including channel fading. We provide performance benchmark for both basic likelihood ratio test and generalized likelihood ratio test under different deployment conditions. Our results clearly demonstrate gain provided by increasing number of collaborative sensors.

Packet-switched wireless networks such as 4G and 5G cellular systems apply RObust Header Compression (ROHC) to reduce PDCP header length and improve payload efficiency. Our recent works have demonstrated the benefit of applying a trans-layer approach that exploits lower layer information in ROHC control based on a partially observable Markov decision process (POMDP) formulation. The benefit of the POMDP solution comes at significant computation complexity beyond existing ROHC. The present work focuses on simplicity by designing ROHC compressors with lower layer awareness including channel adaptive transport block size as a result of link adaptation present in many wireless networks. Our new models directly address practical implementation and can deliver transmission efficiency close to an optimized POMDP compressor.

In many packet-switched wireless systems including cellular networks, RObust Header Compression (ROHC) plays an important role in improving payload efficiency by reducing the number of header bits in a link session. However, there are only very few research works addressing the optimized control of ROHC. Our recent studies have demonstrated the advantage of a trans-layer ROHC design that exploits lower layer link status. We have presented a unidirectional ROHC design based on a partially observable Markov decision process formulation that enables the transmitter to decide the header compression level without receiver feedback. The present work considers the physical channel dynamics in an LTE environment and how they affect header decompressor status. Our new model takes into consideration the transport block (TBs) size defined in LTE transmission according to the modulation and coding scheme (MCS). Our novel and practical model can significantly improve the efficiency of the transmission when compared to a traditional timer-based ROHC control.