


摘 要: 隨著無線網絡不斷增長的業務需求,蜂窩架構頻譜資源受限,回程容量將成為系統瓶頸。為了緩解這種瓶頸,考慮一種特殊的異構蜂窩網絡,結合緩存節點的部署、用戶位置分布、用戶對請求內容的偏好以及緩存節點有限的存儲空間,對內容存儲及用戶關聯聯合優化問題進行建模分析。將目標函數建模為請求時延的最小化,簡單證明該問題是NP-hard的,并設計了基于改進KM(Kuhn-Munkres)的內容放置策略。最后,通過實驗比較了該算法與其他基準方案的性能。
關鍵詞: 無線緩存; 用戶偏好; 緩存時延; 緩存放置
中圖分類號: TP929.5"" 文獻標志碼: A
文章編號: 1001-3695(2022)01-022-0123-05
doi:10.19734/j.issn.1001-3695.2021.05.0215
Collaborative content caching strategy based on user preferences
Zuo Yabing, Wang Kai, Yang Fan, Jiang Jing
(School of Communication amp; Information Engineering, Xi’an University of Posts amp; Telecommunications, Xi’an 710121, China)
Abstract: With the ever-increasing service requirements of wireless networks, the spectrum resources of the cellular architecture are limited, and the backhaul capacity will become a system bottleneck. In order to alleviate this bottleneck, this paper considered a special heterogeneous cellular network, combined with the deployment of cache nodes, user location distribution, and considers user preferences for content requests and the limited storage space of cache nodes, modeling and analysis of the joint optimization problem of content storage and user association. And it modeled the objective function as the minimization of request delay. It simply proves that the problem is NP-hard, and designed a content placement strategy based on improved Kuhn-Munkres. Finally, the performance of this algorithm was compared with other benchmark schemes through experi-ments.
Key words: wireless cache; user preference; cache latency; cache placement
0 引言
隨著近些年具有先進多媒體功能的智能移動設備和高數據速率應用越來越普及,無線網絡的數據流量正經歷著巨大的增長,特別是多媒體數據流[1]。與此同時,移動用戶也越來越需要高質量的多媒體內容,這通常需要額外的網絡資源來分發內容。因此,就內容規模和多樣性而言,需要實時交付給移動用戶的內容需求已增長到前所未有的水平[2,3],宏蜂窩體系結構即使在分配了新的蜂窩頻譜后也無法支持這種業務量的增加。因此本文考慮一種新型異構微小區架構(FemtoCaching),如文獻[4]所總結的,這種架構中有許多小單元接入點,它們具備低回程高存儲的特點。這些具有大存儲容量和僅無線連接的特殊節點(以后稱為FAP)可以部署在小區中,回程僅用于刷新隨時間和用戶需求分布變化的緩存。整個系統的設計和優化復雜,涉及多個層面的研究問題,遠遠超出了本文范圍,但在這里只關注特定的方面,即緩存內容和緩存放置問題。目前,國內外眾多學者對啟用緩存的移動邊緣網絡進行了大量的研究工作,為了在有限緩存空間下為用戶提供更好的服務體驗,眾多有價值的解決方案被研究者提出。首先,解決緩存內容選取問題最常用的方法就是預測內容的流行度,依據內容流行度選擇緩存內容。事實證明,在邊緣緩存站緩存最受歡迎的文件可以有效減少回程流量并提高用戶的服務質量(QoE)[5]。然而,在海量的內容中預測和選擇少數用戶需要的流行內容具有較大的難度,故內容流行度的預測既是邊緣緩存策略的重點也是難點。文獻[6]通過追蹤YouTube網站的用戶觀看記錄發現,視頻內容的流行度非常符合Zipf分布。當前許多研究也都是假設內容的流行度是滿足Zipf分布的,實際上文件流行度是不可能提前知道的。另一方面,根據用戶偏好進行緩存可以獲得更好的緩存性能,因此更有意義的反饋是用戶在消費文件后是否喜歡該文件[7]。
為了解決緩存內容的放置問題,考慮到在傳統的緩存策略中,各基站只存儲本地最流行的內容可能導致緩存冗余以及存儲資源浪費的問題。文獻[8]提出了一種基于分組的分層協同緩存策略以支持本地用戶的請求。文獻[9]基于Lyapunov技術提出了一種新穎的隨機協作內容放置策略,該策略僅使用當前可用的信息作出內容放置決策,利用編碼緩存的空間內容流行度變化,實現了可證明的接近最佳的長期緩存性能,但對未來的內容到達未做預測。文獻[10~12]以最小化能量效率和網絡成本為優化目標進行內容的分發和聯合緩存放置。文獻[13]研究了一種基于馬爾可夫決策過程的組播調度策略,該策略在一定程度上提升了動態內容的傳輸性能。
綜上所述,盡管目前對異構無線網中內容緩存算法的研究有很多,但仍然存在一些問題,不合理的內容存儲及資源分配可能導致個別接入點服務的用戶數過多、資源不足,同時后續用戶難以尋找到合適的緩存及無線資源,這嚴重影響了用戶的服務質量。針對以上研究的不足,本文將結合異構Femtocaching蜂窩網絡中FAP的部署、用戶位置分布,對內容存儲及用戶關聯聯合優化問題進行分析建模,具體貢獻如下:
a)為了增加內容多樣性以及提高緩存命中率,本文在FAP上除了緩存請求率高的流行內容外,還考慮結合用戶偏好,利用基于協同過濾的算法預測出用戶有可能請求的內容。
b)在FemtoCaching協作緩存架構下,將目標函數建模為請求時延的最小化,并簡單證明該問題是NP-hard的。接著考慮區域內各FAP的位置信息和用戶需求的差異,設計了基于改進KM(Kuhn-Munkres)的內容放置策略。最后仿真結果表明,所提策略與基準策略相比,在請求命中率和緩存時延上均有一定程度的提升。
5 結束語
本文研究了FemtoCaching系統中的無線緩存問題,考慮用戶對內容請求的偏好以及FAP上有限的緩存空間,對內容存儲及用戶關聯聯合優化問題進行建模分析,將目標函數建模為請求時延的最小化,本文算法的關鍵思想是基于用戶和緩存助手的位置以及緩存助手當前可用的緩存空間作出決策,從長遠的角度來提高緩存性能。仿真結果表明,所提緩存系統與基準緩存方案相比,性能有較大的提升。下一步考慮針對無線緩存中的內容編碼問題進行分析,以進一步提升緩存系統的性能。
參考文獻:
[1]Sheraz M, Ahmed M, Hou Xueshi, et al. Artificial intelligence for wireless caching: schemes, performance, and challenges[J].IEEE Communications Surveys amp; Tutorials,2021,23(1):631-661.
[2]Zhang Meng, Luo Hongbin, Zhang Hongke. A survey of caching mechanisms in information-centric networking[J].IEEE Communications Surveys amp; Tutorials,2015,17(3):1473-1499.
[3]Kwak J, Kim Y, Le L B, et al. Hybrid content caching in 5G wireless networks:cloud versus edge caching[J].IEEE Trans on Wireless Communications,2018,17(5):3030-3045.
[4]Golrezaei N, Molisch A F, Dimakis A G, et al. Femto-caching and device-to-device collaboration: a new architecture for wireless video distribution[J].IEEE Communications Magazine,2013,51(4):142-149.
[5]Liu Dong, Chen Binqiang, Yang Chenyang, et al. Caching at the wireless edge: design aspects, challenges, and future directions[J].IEEE Communications Magazine,2016,54(9):22-28.
[6]Li Chenyu, Liu Jun. Large-scale characterization of comprehensive online video service in mobile network[C]//Proc of IEEE International Conference on Communications.Piscataway,NJ:IEEE Press,2016:1-7.
[7]Huang Zhiming, Hu Bingshan, Pan Jianping. Caching by user prefe-rence with delayed feedback for heterogeneous cellular networks[J].IEEE Trans on Wireless Communications,2021,20(3):1655-1667.
[8]Ren Dewang, Gui Xiaolin, Lu Wei, et al. GHCC: grouping-based and hierarchical collaborative caching for mobile edge computing[C]//Proc of the 16th International Symposium on Modeling and Optimization in Mobile, Ad hoc, and Wireless Networks.Piscataway,NJ:IEEE Press,2018:1-6.
[9]Chen Lixing, Song Linqi, Chakareski J, et al. Collaborative content placement among wireless edge caching sations with time-to-live cache[J].IEEE Trans on Multimedia,2020,22(2):432-444.
[10]寇蘭,劉彪,黃宏程.用戶需求感知的D2D視頻分發機制[J].計算機應用研究,2019,36(6):1849-1853.(Kou Lan, Liu Biao, Huang Hongcheng. A demand-aware D2D video distribution mechanism[J].Application Research of Computers,2019,36(6):1849-1853.)
[11]Gregori M, Gómez-Vilardebó J, Matamoros J, et al. Wireless content caching for small cell and D2D networks[J].IEEE Journal on Selected Areas in Communications,2016,34(5):1222-1234.
[12]Tadrous J, Eryilmaz A. On optimal proactive caching for mobile networks with demand uncertainties[J].IEEE/ACM Trans on Networking,2015,24(5):2715-2727.
[13]Zhou Bo, Cui Ying, Tao Meixia. Stochastic content-centric multicast scheduling for cache-enabled heterogeneous cellular networks[J].IEEE" """"Trans on Wireless Communications,2016,15(9):6284-6297.
[14]Golrezaei N, Shanmugam K, Dimakis A G, et al. Femtocaching: wireless video content delivery through distributed caching helpers[C]//Proc of IEEE INFOCOM.Piscataway,NJ:IEEE Press,2012:1107-1115.
[15]Li Chenglin, Toni L, Zou Junni, et al. QOE-driven mobile edge ca-ching placement for adaptive video streaming[J].IEEE Trans on Multimedia,2018,20(4):965-984.
[16]Jiang Wei, Feng Gang, Qin Shuang. Optimal cooperative content caching and delivery policy for heterogeneous cellular networks[J].IEEE Trans on Mobile Computing,2017,16(5):1382-1393.
[17]Kuhn H W. The Hungarian method for the assignment problem[J].Naval Research Logistics,2005,52(1):7-21.
[18]Konstan J A, Miller B N, Maltz D, et al. GroupLens: applying collaborative filtering to Usenet news[J].Communications of the ACM,1997,40(3):77-87.
[19]趙向宇.Top-N協同過濾推薦技術研究[D].北京:北京理工大學,2014.(Zhao Xiangyu. Research on top-N collaborative filtering recommendation technology[D].Beijing:Beijing Institute of Techno-logy,2014.)
[20]Resnick, P, Iacovou N, Suchak M, et al. GroupLens: an open architecture for collaborative filtering of netnews [C]//Proc of ACM Conference on Computer Supported Cooperative Work.1994:175-186.
[21]Zhu Haibin, Zhou Mengchu, Alkins R. Group role assignment via a Kuhn-Munkres algorithm-based solution[J].IEEE Trans on Systems, Man, and Cybernetics-Part A:Systems and Humans,2012,42(3):739-750.
[22]Wang Yunting, Xu Xiaodong, Jin Yaqi. QoS constraint optimal load balancing for heterogeneous ultra-dense networks[C]//Proc of the 19th International Symposium on Wireless Personal Multimedia Communications. Piscataway,NJ:IEEE Press, 2016:317-323.
[23]Ugwuanyi E E, Iqbal M, Dagiuklas T. A novel predictive-collaborative-replacement (PCR) intelligent caching scheme for multi-access edge computing[J].IEEE Access,2021,9:37103-37115.