This paper addresses the problem of joint inference and optimization in wireless networks. An optimization framework based on information-geometric network inference is developed and implemented (using a real radio emulation testbed) with scalable solutions to infer the end-to-end rate distributions of stochastic network flows from link rate measurements. The proposed low-complexity solutions apply when the underlying network inference (network tomography) problem can be decomposed to smaller-size subproblems that are solved independently by partially inferring only the flow rates of interest. The solutions are extended to infer flow rates jointly with link loss rates when retransmissions are considered over unreliable wireless links. By using the inferred distributions of flow rates, an inference mechanism is presented to optimize the network performance. First, the distributions of flow rates are inferred from the average rate measurements on selected links. Then, the distributions of all links rates are computed from the inferred distributions of flow rates. Finally, these inference results are used with network optimization. The weighted sum of link outage probabilities is minimized by adapting power control or routing decisions in mobile wireless access. This approach is iterated between network inference and optimization, providing link outage and end-to-end throughput gains compared to static approaches with fixed network (inference and optimization) parameters. The joint network inference and optimization framework is implemented with real configurable radios and the performance is verified with hardware-in-the-loop emulation test results that are obtained with actual radio transmissions over emulated channels.

We study from a network-layer perspective a simple cognitive network consisting of one primary user (PU) and one secondary user (SU) sharing the spectrum with the primary. We propose and analyze two different access schemes at the SU aiming at maximizing its stable throughput while guaranteeing the stability of the PU's queue. These schemes exploit the SU's knowledge of the statistics of the channels as well as the average arrival rate to the PU. The first scheme is where the SU does not perform any sensing but accesses the channel at all slots with fixed probability p*. The second scheme is where the SU senses the channel at all slots and accesses the channel with probabilities p1* and p2* when the PU is sensed to be idle and busy respectively. We also compare these schemes to the traditional opportunistic spectrum access (OSA) where the SU accesses the channel with probability one only if the channel is sensed to be idle. The analysis shows that if the PU and/or SU receivers can decode simultaneous transmissions with high success probability, then schemes with no sensing are preferred as they provide to the SU more duration for data transmission and they outperform schemes with sensing. In this case, the OSA scheme is over- protective. Otherwise, schemes with sensing are preferred since sensing is crucial for PU protection. Therefore, using complex receivers that can handle simultaneous transmissions successfully alleviates the need of complex SU transmitters with strong sensing capability which might be preferred in some circumstances.

We consider the problem of reliable spectrum sensing and opportunistic access on channels with stochastic traffic in batch processing systems such as network coding (NC). We show how a secondary user (SU) can leverage the structure induced by block-based NC on primary users' (PUs) channels to mitigate the effects of channel sensing errors and improve the detection of idle PU spectrum and the throughput. NC is known to improve the transmission efficiency and therefore when applied on a PU channel, it can extend the spectrum availability for the SUs. We refer to the additional gain of spectrum predictability from NC and show that under possible sensing errors the SU can more reliably detect the idle spectrum if the PUs' channels carry network-coded transmissions even when the channel utilization is fixed. We consider two different objectives at the SU. For quickest detection, the SU applies the Cumulative Summation (CUSUM) algorithm to detect idle slots on a PU channel and further improves the detection capability with the Viterbi algorithm, if the PU spectrum dynamics are known. For throughput maximization, the SU tracks the PU spectrum with the Partially Observable Markov Decision Process (POMDP) approach. Our results show that NC renders the spectrum more predictable, which can be used by the SUs to mitigate the effects of sensing errors and improve the throughput. We validate these results with real radio measurements taken in software-defined radio based wireless network tests.

We study from a cross-layer perspective a cognitive network consisting of one primary user (PU) and one secondary user (SU). In contrast with the oversimplified collision channel model, we assume that simultaneous PU and SU transmissions are successful with some positive probability. We propose and analyze two access schemes at the SU aiming at maximizing its stable throughput while guaranteeing the stability of the PU's queue. These schemes exploit the SU's knowledge of the channel statistics and of the average arrival rate to the PU. In the first scheme, the SU accesses the channel at all slots with fixed probability p^* without sensing. In the second scheme, the SU accesses the channel with probabilities p_1^* and p_2^* when the PU is sensed to be idle and busy, respectively. The analysis shows that if simultaneous PU and SU transmissions are likely to be successful, schemes with no sensing outperform schemes with sensing. Otherwise, schemes with sensing are preferred. Therefore, using complex receivers capable of handling interference, alleviates the need of complex SU transmitters with strong sensing capabilities. We then extend the analysis to the case where the PU has an average delay constraint which is more restrictive than the stability constraint.

We consider a secondary node with fixed energy budget per frame aiming at exploiting the time intervals of primary inactivity. Although a long sensing duration leads to better primary protection, it reduces the time and energy left for the secondary transmission. We seek to exploit the knowledge of the cross channel state information (CCSI) between the secondary transmitter and the primary receiver in order to minimize the required sensing duration, thereby increasing the secondary time and energy for transmission leading to an increase in his throughput. The optimal sensing duration is derived and the benefit of exploiting the cross channel state information is substantiated.

We study, from a network layer perspective, the effect of an Ad-Hoc secondary network with N nodes accessing the spectrum licensed to a primary node. If the sensing is perfect, then the secondary nodes do not interfere with the primary node and thus do not affect its stable throughput. In case of imperfect sensing, it is shown that if the primary node's arrival rate is less than some calculated value, then the secondary transmissions do not affect its queueing stability; otherwise, the secondary nodes should regulate their transmission parameters to reduce their interference on the primary. It is also shown that in contrast with the primary user's maximum stable throughput rate which strictly decreases with increased sensing errors, the throughput of the secondary nodes might increase with sensing errors as more transmission opportunities become available to them. Finally, we explore the use of the secondary nodes as relays of the primary node's traffic to compensate for the interference they might cause. In this case, for appropriate modulation scheme and under perfect sensing, it is shown that the more secondary nodes in the system, the better for the primary user in terms of his stable throughput. Meanwhile, the secondary nodes might benefit from relaying by having access to a larger number of idle slots becoming available to them due to the increase of the service rate of the primary. For the case of a single secondary node, the proposed relaying protocol guarantees that either both the primary and the secondary benefit from relaying or none of them does.

We consider a cognitive network consisting of one primary source-destination pair and N secondary cognitive source-destination pairs that randomly access the channel during the primary user's idle slots. We first study the effect of the secondary nodes' transmission parameters such as power and channel access probabilities on the stable throughput of the primary node. If the sensing is perfect, then the secondary nodes do not interfere with the primary node and thus do not affect its stable throughput. In case of imperfect sensing, it is shown that if the primary node's arrival rate is less than some calculated value, then the secondary transmissions do not affect its queueing stability; otherwise, the secondary nodes should regulate their transmission parameters to reduce their interference on the primary. Finally, we propose a multinode relaying protocol based on distributed space-time orthogonal block codes, that uses the secondary nodes as relays of the primary node's traffic to compensate for the interference they might cause. In this case, for appropriate modulation scheme and under perfect sensing, it is shown that the more secondary nodes in the system, the better for the primary user in terms of his stable throughput. Meanwhile, the secondary nodes might benefit from relaying by having access to a larger number of idle slots becoming available to them due to the increase of the service rate of the primary.

In this paper, we consider the problem of calculating the stability region and average delay of two user slotted ALOHA over a Gilbert-Elliott channel, where users have channel state information and adapt their transmission probabilities according to the channel state. Each channel has two states, namely, the `good' and `bad' states. In the `bad' state, the channel is assumed to be in deep fade and the transmission fails with probability one, while in the `good' state, there is some positive success probability. We calculate the stability region with and without Multipacket Reception capability as well as the average delay without MPR. Our results show that the stability region of the controlled S-ALOHA is always a superset of the stability region of uncontrolled S-ALOHA. Moreover, if the channel tends to be in the `bad' state for long proportion of time, then the stability region is a convex polygon strictly containing the TDMA stability region and the optimal transmission strategy is to transmit with probability one whenever the nodes have packets and it is shown that this strategy is delay optimal. On the other hand, if the channel tends to be in the `good' state more often, then the boundary of the stability region is characterized by a convex curve and is strict subset of the TDMA stability region. We also show that enhancing the physical layer by allowing MPR capability can significantly enhance the performance while simplifying the MAC Layer design by the lack of the need of scheduling under some conditions. Furthermore, it is shown that transmission control not only allows handling higher stable arrival rates but also leads to lower delay for the same arrival rate compared with ordinary S-ALOHA.

In this paper, we investigate the impact of network coding at the relay node on the stable throughput rate in multicasting cooperative wireless networks. The proposed protocol adopts Network-level cooperation, as in contrast to the traditional physical layer cooperative protocols; and in addition, it uses random linear network coding at the relay node. The traffic is assumed to be bursty and the relay node forwards its packets during the idle periods of the source which allows better utilization of channel resources. Our results show that cooperation leads to higher stable throughput rates than conventional retransmission policies. Moreover, the use of random linear network coding at the relay can further enhance the stable throughput with increasing network coding field size or increasing the number of packets over which encoding is performed.