چكيده به لاتين
The sixth generation wireless cellular network is considered to be a unique development compared to previous generations of wireless communication systems. Due to the lack of radio resources and the limited capacity of the backhaul links, it is increasingly impossible for all subscribers to access the data they want from remote cloud networks. To meet these unprecedented challenges, network edge caching has been identified as a promising solution. Caching of content on wireless devices in conjunction with device-to-device (D2D) communications allows to exploit this property, and provide a network throughput that is significantly in excess of both the conventional approach of unicasting from cellular base stations and the traditional D2D networks for regular data traffic.
Many content caching studies have assumed that information is fully accessible (For example: channel state information, the popularity of the content, the movement of users, the entry and exit of users, the rate of demand for receiving content, etc). In practice, due to the dynamic conditions and random nature of wireless channels and other parameters, the information obtained are uncertain. To face the challenge of uncertainty, an effective approach is to use robust optimization techniques in wireless communication. In robust optimization, the goal is to make a decision that is feasible no matter what the constraints turn out to be, and optimal for the worst-case objective function. In this thesis, we have considered the uncertainty in the popularity of contents. Given that popular content is measured, this measurement may be erroneous due to the uncertainty and we have to actually pay attention to this amount of error. Few jobs that have considered uncertainty have used the most conservative method of robust optimization. But our method is to use the Bertsimas and Sim method. In this way, the measured popular content has only a certain number (gamma) of uncertainty and the rest of their value is definite. For those whose value is not definite, we consider the worst case.
In the following, we examine the robust method in different scenarios, ie the effect of increasing and decreasing in parameters such as gamma, popularity, deviation, network topology, and so on. As the gamma increases, the value of the objective function decreases by about 8%. In the case of message popularity, as the amount of popularity increases, the value of the target function also increases depending on the amount of popularity. By increasing the amount of deviation of the target function decreases by about 4% and also depending on the number of devices, the amount of the objective function increases. Finally, we will show that if we anticipate a deviation, eventually the actual value of the objective function will increase by an equal amount, or even in some cases by about 3.33%, during network execution.
