QoE-based Log Collection in Minimization of Drive Tests

摘要

We have been conducting research into the creation of a coverage map or a map of coverage holes for mobile access systems, such as cellular systems, by collecting coverage information that includes radio signal information and location information from mobile terminals. A similar study started in 3GPP, known as MDT (Minimization of Drive Tests). In such studies, poorer radio measurements than a pre-defined threshold are one of triggers to collect coverage information. However, it is not sufficient to collect this information based only on the radio measurements criteria since it would be hard to detect an event of quality degradation that is caused, for example, by mobile traffic congestion, which is also a target event that the coverage map should include and against which telecom operators take necessary measures such as certain parameter tunings at base stations and deployment of new base stations. In this paper, we propose a methoa in which a mobile terminal records traffic behavior, such as an increasing rate of received traffic volume and the ratio of downlink packets to uplink packets, per application used, and deduces the QoE (Quality of Experience) of a new communication session based on the recorded traffic behaviors. The information collection trigger based on the proposed method would enable operators to gather information in the event of user experience deterioration, for instance, slower web response than usual. We executed preliminary evaluations in which two research participants used smartphones where the proposed method was implemented, and found that the method correctly judged sessions of poor QoE as being of poor qualities in 75 percent probability.%筆者らは，セルラシステムなどの無線システムのカバレッジマップ/カバレッジホールマップを，受信信号強度や受信信号品質といつた無線品質情報と位置情報とから構成される移動端末からの測定レポートにより構築する検討を行つており,3GPPにおいてもMDT(Minimization of Drive Tests)と呼ばれる同様の取り組みがなされている.それらの取り組みでは，受信信号強度や受信信号品質が所定閾値以下になる事象を測定レポートの生成トリガにすることを想定している.ここで，当該マップの用途としては，無線サービス品質向上を目的とした力バレッジ最適化や基地局増設等であると考えられるが，前述の受信信号強度や受信信号品質といった無線信号情報のみに基づいたレポートトリガでは，例えばトラヒック集中による通信品質低下を検出することが難しく，無線サービス品質向上に資する上で十分とは言えない.そこで，本稿では，利用されるアプリケーション毎のトラヒック利用状況（受信トラヒック量の変動や上り方向／下り方向のパケット数比率等）を移動端末内で学習し，当該学習結果を元に新たに発生したトラヒックの品質に対する良否を推定する手法を提案する.本手法に基づく測定レポ一トトリガにより，（無線品質は良好だが）いつもよりもレスポンスが遅い，といった主観的な通信品質（QoE; Quality of Experience)低下事象の検出が可能になることが期待される.提案手法を実装したスマートフォンの被験者利用に基 づく予備評価の結果，QoE低下事象を75%の割合で検出できた.