Description

I. INTRODUCTION
The drastically increasing multimedia content consumption by the Internet leads to high network traffic. This necessitates the improvement of system capacity and alleviation of latency. Simultaneously, the system energy compensation requirement arises. The content request volume for video content traffic emerges content-centric design of networks to resolve these technical issues [1]. The classical networking techniques are revised for improving the content dissemination in terms of the capacity, latency and energy. The Device-to-device (D2D) technique is a promising enabler for boosting the system capacity while with the short distance transmission technique pursues the energy efficiency [2]. The cognitive radio technique also has a potential for the capacity improvement in 5G networks [3]. Information-Centric Network (ICN) directly requests multimedia from the network instead of the classical IP-based approach [4]. Thus, the focal is put onto the multimedia content. For the substantial system capacity expansion and energy expenditure reduction motives, the layering concept for video contents can be utilized [5]. The content layering improves the scalability [6]. Some content layers are more lucrative in terms of magnifying the network gains. For instance, the base layer is essential since without it videos cannot be displayed [7], [8]. From the content-convergent network orientation aspect, the amalgamation of cognitive radio, D2D paradigm for the dissemination of layered multimedia contents is functional.
Caching is a fundamental capability for improving the system energy expenditure, reducing the latency and amplifying the capacity. Thereof, you will focus on the caching aspect and propose your own caching technique in D2D networks for layered multimedia content transmission. You will implement the Least Recently Used (LRU) and Least Frequently Used (LFU) techniques as baseline profiles. Then you will propose your own caching technique and compare with the baseline profiles.
II. SYSTEM MODEL
In this section, we give the model of our network system. In our system environment, users are dispersed in the spatial domain. For the user locality Poisson Point Process (PPP) is a commonly utilized distribution [9], [10]. In our system, users are distributed according to PPP with mean density in a region with the radius R = 300m. Users hold devices with the repositories that are capable of storing contents. These devices can share video content with each other with the help of D2D paradigm. When a content is requested, first the requester will check its local cache. If not found, it will try to get the service in D2D operation mode. It will fetch the requested content from the closest device that

Fig. 1: Network Structure

Fig. 2: Layered content model.
stores the requested video content. The failed requests are not stored in a buffer for prospective retrials.
The content layering is used for the sake of temporal and spatial scalability [11]. The content layers are named as i) base and ii) enhancement. The base layer is the standard quality part. The enhancement layer is the extra bits requires for high quality viewability. Without the base layer, enhancement layer is of no use. i.e. You can not view the video content. In our system 100 distinct contents exist. The popularity of videos contents vary. Consider the Youtube videos. The trending videos are requestly widely whereas some videos are rarely viewed. For the condensation of the popularity characterization Zipf distribution is utilized [12], [13]. The video popularities follow the Zipf distribution with parameter s = 0.8. Say there are N=100 many different contents, then the popularity of the kth content is . Each content in our system consists of the aforementioned layers. The mean base content size is 25 Mbits while the mean enhancement content size is 5 Mbits [5]. Both layer content sizes follow exponential distribution with the given mean values. The cache capacities of user devices are 1 Gbits. Initially you fill the caches of all users with contents based on their popularities. You can propose your own cache reservation mechanisms for different content layers. This is your design choice that you need to explicitly explain in your project report.
Our devices operate in the terrestrial channel opportunistically (cognitive radio users/secondary users). i.e. The channel belongs to other users (primary users). If these users are active in some frequency our users cannot access the network at that frequency. You can think of the primary users as the house owner of a frequency and our cognitive (secondary) users as the renter of the frequency. Note that in the cognitive radio when a primary user (PU) starts transmission in a frequency the secondary user (SU) should preempt the frequency immediately. You can assume perfect sensing, so no collisions of PU and SUs occur. There exist 10 terrestrial frequencies each of bandwidth B = 2 MHz. The initial frequency f1 is at 700 MHz. The second frequency f2 is at 700+2 MHz. Continuing this way ith frequency is calculated according to the formula 700 + (i − 1) · 2 MHz. You will calculate the capacity of any frequency fi under additive White Gaussian Noise with
Shannon’s capacity formula . Here
B = 2 MHz, N0 = 1.6e−19 WattHz and , where d is the distance between the transmitter and the receiver of the D2D operation. For primary users you can take
. The user arrivals into the system are exponentially distributed with mean for primary users and for our secondary users. Parametrize user request rates and other given parameters (R,s, mean content sizes etc.) so that in need of a tune operation you will not lose time. Create a parameter list function and read all the system parameters from that function. An example channel history is given in the Fig. 3. Note that a frequency can be used by one service at a time. You can assume that PUs request only base layer contents. The rate of SUs that request both base and enhancement layer is a system parameter pHQ = 0.5.
Say a PU request arrives for the content service environment at frequency fi, i ∈ {1,2,…,10} with rate .
• If that frequency fi is idle (empty), just start the service. • Else if frequency fi is not idle but a PU exists, then PU is blocked.
• Else if that frequency fi is not idle but a SU exists, cut the SU operation (base or enhancement) at that frequency and start the PU operation. For the ceased SU:
– If at least one idle frequency exists at the time being, then continue the operation from some idle frequency (shown by light orange in Fig. 3).
– If no idle frequency exists for the continuation of the SU operation the SU is dropped (shown by dark orange in Fig. 3).
This is how you will process the PU arrivals to the system. For the SUs the arrival processing mechanism is as follows:
• If a SU request arrives to the content service environment a random idle frequency will be selected for the service.
• If no idle frequency exists that request is blocked.
• Else if some idle frequency is found and a SU requests
– base layer only then start service at that idle frequency
– both base and enhancement layers, you can use two different techniques:
∗ First serve the base layer and after successful completion of the base layer transmission start serving the enhancement layer immediately at the same frequency as shown for f2 in Fig. 3.
∗ Start the services for both base and enhancement components simultaneously at different frequencies shown in the first services for f3 and f4 in Fig. 3.
You need to keep track of channel status and log service, block and drop events of PUs (of type base) and SUs (of type base and enhancement). Think of your data structures for the channel status and events.

Fig. 3: Channel history example.
III. CACHING ALGORITHMS
Let us briefly summarize what you will be doing in your project.
• Investigate current literature to learn about key concepts and the state-of-the-art for caching mainly the papers [14,15,16,17,18,19].
• Propose your own caching technique and tell how it differs from the papers [14,15,16,17,18,19].
• Do the time complexity analysis of your caching technique
• Identify data structures that you will use in your project for channel management
• Implement your caching algorithm, LRU and LFU
• Visualize cache hit rate and average latency values of your caching algorithm, LRU and LFU
• Attend the demo session and show your code and results
REFERENCES
[1] G. Gur, “Energy-aware cache management at the wireless network edge¨ for information-centric operation,” Journal of Network and Computer Applications, vol. 57, pp. 33 – 42, 2015.
[3] C. Wang, F. Haider, X. Gao, X. You, Y. Yang, D. Yuan, H. M.
[6] J. . Ohm, “Advances in scalable video coding,” Proceedings of the IEEE, vol. 93, no. 1, pp. 42–56, Jan 2005.
[7] S. Ullah, T. LeAnh, A. Ndikumana, M. G. R. Alam, and C. S. Hong, “Layered video communication in icn enabled cellular network with d2d communication,” in 2017 19th Asia-Pacific Network Operations and Management Symposium (APNOMS), Sept 2017, pp. 199–204.
[8] J. . Ohm, “Advances in scalable video coding,” Proceedings of the IEEE, vol. 93, no. 1, pp. 42–56, Jan 2005.
[9] D. Malak and M. Al-Shalash, “Optimal caching for device-to-device content distribution in 5g networks,” in 2014 IEEE Globecom Workshops (GC Wkshps), Dec 2014, pp. 863–868.
[11] P. Seeling, M. Reisslein, and B. Kulapala, “Network performance evaluation using frame size and quality traces of single-layer and twolayer video: A tutorial,” IEEE Communications Surveys Tutorials, vol. 6, no. 3, pp. 58–78, Third 2004.
[12] D. Malak and M. Al-Shalash, “Optimal caching for device-to-device content distribution in 5g networks,” in 2014 IEEE Globecom Workshops (GC Wkshps), Dec 2014, pp. 863–868.
[13] M. Emara, H. ElSawy, S. Sorour, S. Al-Ghadhban, M. Alouini, and T. Y. Al-Naffouri, “Optimal caching in multicast 5g networks with opportunistic spectrum access,” in GLOBECOM 2017 – 2017 IEEE Global Communications Conference, Dec 2017, pp. 1–7.
[14] I. Psaras, W. K. Chai, and G. Pavlou, “Probabilistic in-network caching for information-centric networks,” in Proceedings of the Second Edition of the ICN Workshop on Information-centric Networking, ser. ICN ’12. New York, NY, USA: ACM, 2012, pp. 55–60. [Online]. Available:
http://doi.acm.org/10.1145/2342488.2342501
[15] D. Hong, D. De Vleeschauwer, and F. Baccelli, “A chunk-based caching algorithm for streaming video,” in NET-COOP 2010 – 4th Workshop on Network Control and Optimization, Gent, Belgium, Nov. 2010, session 05 : Streaming applications.
[16] K. Suksomboon, S. Tarnoi, Y. Ji, M. Koibuchi, K. Fukuda, S. Abe, N. Motonori, M. Aoki, S. Urushidani, and S. Yamada, “Popcache: Cache more or less based on content popularity for information-centric networking,” in 38th Annual IEEE Conference on Local Computer Networks, Oct 2013, pp. 236–243.
[19] H. Zhu, Y. Cao, W. Wang, B. Liu, and T. Jiang, “Qoe-aware resource allocation for adaptive device-to-device video streaming,” IEEE Network, vol. 29, no. 6, pp. 6–12, Nov 2015.