Co-existence and cooperation in dynamic shared spectrum networks

Wireless networks that share the spectrum using dynamic spectrum access techniques have been proposed as a solution to the spectrum scarcity problem created due to huge increases in the number of mobile application and devices. The main difference between these networks and traditional wireless networks is in the heterogeneous nature of channel access. Namely, unlicensed users, also known as secondary users, in such networks have lower priority than licensed, also known as primary users. In this work we look at some problems of coexistence and cooperation between primary and secondary networks.;First, we study cooperative scheduling in a network with a single macro primary base station and multiple small secondary base stations in which the secondary base station can relay primary users packets. Under this cooperative scheme, we investigate whether, and under what conditions, the primary and secondary networks can be stabilized without explicit knowledge of the packet arrival rates. In particular, we consider a deterministic primary packet generation process but random secondary packet generation process. This packet generation process models a real-time Constant Bit Rate (CBR) voice or video source. For this network we develop a relaying and scheduling algorithm using Lyapunov drift techniques that does not require knowledge of packet arrival rates. We obtain a guaranteed stability region, consisting of packet generation rates for which the network can be stabilized under this algorithm. From this region we observe that when the primary packet generation rate is lower than what can be maximally supported without cooperation, the set of secondary packet arrival rate vectors for which the network can be stabilized do not decrease under cooperation. For higher primary packet generation rates the algorithm stabilizes the network for a non-empty set of secondary packet arrival rate vectors.;Second, we characterize the average end-to-end delay in an opportunistic multi-hop ad-hoc secondary network overlaid with a primary multi-hop ad-hoc network, both with and without cooperation between the two networks. Nodes in both networks use random medium access control (MAC) scheme with exponentially distributed back-off. We first consider the non-cooperative case and model the whole network as a two-class priority queuing network and use queuing-theoretic approximation techniques to obtain a set of relations involving the mean and second moments of the inter-arrival time and service-time of packets at a secondary node. Then, applying those parameters to an equivalent open single-class G/G/1-queuing network and using diffusion approximation, we obtain expressions for the average end-to-end delay of a packet in the secondary network. We then extend the analysis to a case where secondary nodes cooperatively relay primary packets so as to improve their own transmission opportunities. The mathematical results are validated against extensive simulations.;Third, we study cooperation between primary and secondary users via caching. We consider a network consisting of a single macro (primary) base-station and multiple small (secondary) base-stations. The secondary base-stations can cache some primary files and thereby serve content requests from nearby primary users. For this cooperative scenario, we develop two caching and scheduling policies under which the set of primary and secondary user request generation rates that can be supported increases from the case without cooperation. The first of these algorithms provides maximum gain in the set of supportable primary and secondary request generation rates. However under this algorithm primary packet transmissions from secondary base-stations do not have higher priority than that of secondary packets. As a result, we propose another sub-optimal (with respect to set of supportable request generation rate vectors) algorithm wherein transmissions of primary packets from secondary base-stations have higher priority than that of secondary packets. We conduct extensive simulations to compare the performance of both algorithms with that of a non-cooperative algorithm that is optimal, with respect to set of supportable request generation rates, among all non-cooperative policies.;Finally, we study proactive scheduling in a network consisting of a single macro base station and multiple secondary small base stations. Unlike the previous three scenarios, in this network both the macro and small base stations are owned by the same operator and serve the same set of users. However, the small base stations act as secondary transmitters and use dynamic spectrum access to avoid interfering with a primary base station transmission. We study the benefit of using proactive scheduling in such networks by comparing the asymptotic values of the average system costs under a proactive policy and that under a reactive policy. We also study the dependence of these costs on channel gains, user request rates, the number of users and base stations.